PREDICCION ACCIDENTES MEDELLIN
Edwin Villarraga
16/7/2019
DESCRIPCION DEL PROBLEMA
HIPOTESIS
CARGA Y DEPURACION DE DATOS
Carga de Datos
Datos de 2014
Durante la carga de datos se realizo la verificacion de las clases de accidentes e inicialmente se identificaron 6 categorias: 1. Atropello 2. Caida_Ocupante 3. Choque 4. Incendio 5. Otro 6. Volcamiento
Sin embargo entre 2014 y 2017 solo se presentaron 17 casos de incendio y en el ano 2018, 7 casos. Dada la poca cantidad de eventos se dejaron finalmente 5 categorias dado que incendio se incorporo a otro:
- Atropello
- Caida_Ocupante
- Choque
- Otro
- Volcamiento
raw_data_2014 <- read.csv(file="./data/Accidentalidad_georreferenciada_2014.csv", encoding="UTF-8", header=TRUE, sep=",")head(raw_data_2014)## X.U.FEFF.OBJECTID X Y RADICADO FECHA
## 1 211279 831190.1 1179690 1423828 2014-01-01T00:00:00.000Z
## 2 211280 835013.6 1184224 1423839 2014-01-01T00:00:00.000Z
## 3 211281 837032.7 1184750 1423840 2014-01-01T00:00:00.000Z
## 4 211282 830649.5 1181383 1423849 2014-01-01T00:00:00.000Z
## 5 211283 833740.8 1188644 1423890 2014-01-01T00:00:00.000Z
## 6 211284 836425.4 1186933 1423892 2014-01-01T00:00:00.000Z
## HORA DIA PERIODO CLASE DIRECCION
## 1 02:20 AM 1 2014 Choque CR 80 CL 8
## 2 12:50 AM 1 2014 Atropello CR 53 CL 61
## 3 01:00 AM 1 2014 Atropello CR 39 CL 70
## 4 12:37 AM 1 2014 Atropello CL 32 CR 84
## 5 10:40 AM 1 2014 Caída de Ocupante CR 80 CL 101
## 6 04:00 AM 1 2014 Choque CR 48 CL 93
## DIRECCION_ENC CBML TIPO_GEOCOD GRAVEDAD
## 1 CR 080 008 000 00000 1611 Malla vial aproximada: CR 81-7 HERIDO
## 2 CR 053 061 000 00000 1003 Malla vial HERIDO
## 3 CR 039 070 000 00000 0308 Malla vial HERIDO
## 4 CL 032 084 000 00000 1617 Malla vial HERIDO
## 5 CR 080 101 000 00000 0603 Malla vial HERIDO
## 6 CR 048 093 000 00000 0401 Malla vial HERIDO
## BARRIO COMUNA DISENO DIA_NOMBRE MES
## 1 Loma de los Bernal Belén Tramo de via MIÉRCOLES 1
## 2 Jesús Nazareno La Candelaria Tramo de via MIÉRCOLES 1
## 3 Manrique Oriental Manrique Tramo de via MIÉRCOLES 1
## 4 Las Mercedes Belén Tramo de via MIÉRCOLES 1
## 5 Doce de Octubre No.2 Doce de Octubre Tramo de via MIÉRCOLES 1
## 6 Berlin Aranjuez Interseccion MIÉRCOLES 1summary(raw_data_2014)## X.U.FEFF.OBJECTID X Y RADICADO
## Min. :211279 Min. :823074 Min. :1172279 Min. :3.800e+01
## 1st Qu.:222939 1st Qu.:833036 1st Qu.:1181405 1st Qu.:1.436e+06
## Median :234284 Median :834331 Median :1183250 Median :1.447e+06
## Mean :234276 Mean :834150 Mean :1183303 Mean :4.810e+15
## 3rd Qu.:245700 3rd Qu.:835417 3rd Qu.:1185597 3rd Qu.:1.459e+06
## Max. :257033 Max. :845503 Max. :1189981 Max. :5.002e+19
##
## FECHA HORA DIA
## 2014-06-03T00:00:00.000Z: 168 04:00 PM: 776 Min. : 1.00
## 2014-02-13T00:00:00.000Z: 163 06:00 PM: 768 1st Qu.: 8.00
## 2014-08-06T00:00:00.000Z: 163 05:00 PM: 766 Median :16.00
## 2014-08-08T00:00:00.000Z: 158 02:00 PM: 705 Mean :15.73
## 2014-05-19T00:00:00.000Z: 157 03:00 PM: 695 3rd Qu.:23.00
## 2014-07-07T00:00:00.000Z: 157 01:00 PM: 679 Max. :31.00
## (Other) :40628 (Other) :37205
## PERIODO CLASE DIRECCION
## Min. :2014 Atropello : 4779 CR 64 C CL 78: 230
## 1st Qu.:2014 Caída de Ocupante: 4157 CR 1 CL 1 : 165
## Median :2014 Choque :27157 CR 64 C CL 67: 151
## Mean :2014 Incendio : 8 CR 57 CL 44 : 118
## 3rd Qu.:2014 Otro : 4521 CR 80 CL 50 : 115
## Max. :2014 Volcamiento : 972 CR 80 CL 65 : 115
## (Other) :40700
## DIRECCION_ENC CBML
## CR 064 C 078 000 00000: 248 1019 : 1148
## CR 001 001 000 00000 : 165 0517 : 979
## CR 064 C 067 000 00000: 151 1507 : 783
## CR 057 044 000 00000 : 118 1012 : 774
## CR 080 065 000 00000 : 116 1007 : 733
## CR 080 050 000 00000 : 115 1105 : 733
## (Other) :40681 (Other):36444
## TIPO_GEOCOD GRAVEDAD
## Malla vial :30320 HERIDO :23077
## Malla vial cruce invertido : 2251 MUERTO : 256
## EPM sin Interior : 987 SOLO DAÑOS:18261
## EPM con Interior : 435
## Malla vial aproximada: CR 1-2: 167
## ZONA RURAL : 97
## (Other) : 7337
## BARRIO COMUNA DISENO
## La Candelaria : 1151 La Candelaria : 8760 Tramo de via :33745
## Caribe : 980 Laureles Estadio: 4678 Interseccion : 5305
## Campo Amor : 786 Castilla : 4204 Lote o Predio: 1083
## Perpetuo Socorro : 775 El Poblado : 3154 Glorieta : 702
## Los Conquistadores: 735 Guayabal : 2946 : 256
## Guayaquil : 733 Robledo : 2828 Ciclo Ruta : 192
## (Other) :36434 (Other) :15024 (Other) : 311
## DIA_NOMBRE MES
## DOMINGO :4209 Min. : 1.000
## JUEVES :6143 1st Qu.: 4.000
## LUNES :6008 Median : 7.000
## MARTES :6301 Mean : 6.543
## MIÉRCOLES:6420 3rd Qu.: 9.000
## SÁBADO :5984 Max. :12.000
## VIERNES :6529str(raw_data_2014)## 'data.frame': 41594 obs. of 19 variables:
## $ X.U.FEFF.OBJECTID: int 211279 211280 211281 211282 211283 211284 211285 211286 211287 211288 ...
## $ X : num 831190 835014 837033 830649 833741 ...
## $ Y : num 1179690 1184224 1184750 1181383 1188644 ...
## $ RADICADO : num 1423828 1423839 1423840 1423849 1423890 ...
## $ FECHA : Factor w/ 365 levels "2014-01-01T00:00:00.000Z",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ HORA : Factor w/ 967 levels "01:00 AM","01:00 PM",..: 95 954 1 938 777 201 469 268 37 72 ...
## $ DIA : int 1 1 1 1 1 1 1 1 1 1 ...
## $ PERIODO : int 2014 2014 2014 2014 2014 2014 2014 2014 2014 2014 ...
## $ CLASE : Factor w/ 6 levels "Atropello","Caída de Ocupante",..: 3 1 1 1 2 3 3 3 3 2 ...
## $ DIRECCION : Factor w/ 12673 levels "CL 1 A CR 42",..: 11441 8871 6987 1424 11297 8063 3081 6269 11142 8131 ...
## $ DIRECCION_ENC : Factor w/ 12510 levels "000","CL 029 000 00000",..: 11024 8642 6786 915 11166 7817 2469 6072 10863 7898 ...
## $ CBML : Factor w/ 391 levels "","0101","01010480009",..: 333 192 39 340 105 48 170 176 241 280 ...
## $ TIPO_GEOCOD : Factor w/ 2252 levels "Catastro con Interior",..: 2064 5 5 5 5 5 5 5 5 5 ...
## $ GRAVEDAD : Factor w/ 3 levels "HERIDO","MUERTO",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ BARRIO : Factor w/ 313 levels "","6001","Aguas Frias",..: 181 129 201 175 81 40 31 155 294 302 ...
## $ COMUNA : Factor w/ 24 levels "","Aranjuez",..: 4 16 18 4 11 2 5 5 17 12 ...
## $ DISENO : Factor w/ 13 levels "","Ciclo Ruta",..: 11 11 11 11 11 4 11 4 11 11 ...
## $ DIA_NOMBRE : Factor w/ 7 levels "DOMINGO ","JUEVES ",..: 5 5 5 5 5 5 5 5 5 5 ...
## $ MES : int 1 1 1 1 1 1 1 1 1 1 ...levels(raw_data_2014$CLASE)## [1] "Atropello" "Caída de Ocupante" "Choque"
## [4] "Incendio" "Otro" "Volcamiento"levels(raw_data_2014$CLASE) <- c("Atropello","Caida_Ocupante","Choque","Otro","Otro",
"Volcamiento")levels(raw_data_2014$CLASE)## [1] "Atropello" "Caida_Ocupante" "Choque" "Otro"
## [5] "Volcamiento"Datos de 2015
raw_data_2015 <- read.csv(file="./data/Accidentalidad_georreferenciada_2015.csv", encoding="UTF-8", header=TRUE, sep=",")head(raw_data_2015)## X.U.FEFF.OBJECTID X Y RADICADO FECHA
## 1 1001 835766.9 1183119 1471803 2015-01-13T00:00:00.000Z
## 2 1002 839496.9 1177763 1471804 2015-01-13T00:00:00.000Z
## 3 1003 832970.8 1178867 1471805 2015-01-13T00:00:00.000Z
## 4 1004 831318.1 1187204 1471806 2015-01-13T00:00:00.000Z
## 5 1005 833710.7 1185631 1471807 2015-01-13T00:00:00.000Z
## 6 1006 839496.9 1177763 1471808 2015-01-12T00:00:00.000Z
## HORA DIA PERIODO CLASE DIRECCION
## 1 02:20 PM 13 2015 Volcamiento CR 43 CL 54
## 2 09:00 AM 13 2015 Otro CR 1 CL 1
## 3 10:00 AM 13 2015 Volcamiento CR 58 D CL 3
## 4 06:30 PM 13 2015 Otro CL 68 B CR 96 E
## 5 05:40 AM 13 2015 Otro CR 68 CL 70
## 6 09:30 AM 12 2015 Volcamiento CR 1 CL 1
## DIRECCION_ENC CBML TIPO_GEOCOD
## 1 CR 043 054 000 00000 1016 Malla vial
## 2 CR 001 001 000 00000 9086 Malla vial aproximada: CR 1-2
## 3 CR 058 D 003 000 00000 1507 Malla vial aproximada: CR 58-4
## 4 CL 068 B 096 E 000 00000 0724 Malla vial aproximada: CL 68-95
## 5 CR 068 070 000 00000 0519 Malla vial aproximada: CR 69-71
## 6 CR 001 001 000 00000 9086 Malla vial aproximada: CR 1-2
## GRAVEDAD BARRIO COMUNA DISENO
## 1 HERIDO Boston La Candelaria Tramo de via
## 2 HERIDO Suburbano El Plan Corregimiento de Santa Elena Tramo de via
## 3 HERIDO Campo Amor Guayabal Tramo de via
## 4 HERIDO Monteclaro Robledo Tramo de via
## 5 HERIDO El Progreso Castilla Tramo de via
## 6 HERIDO Suburbano El Plan Corregimiento de Santa Elena Tramo de via
## DIA_NOMBRE MES
## 1 MARTES 1
## 2 MARTES 1
## 3 MARTES 1
## 4 MARTES 1
## 5 MARTES 1
## 6 LUNES 1summary(raw_data_2015)## X.U.FEFF.OBJECTID X Y RADICADO
## Min. : 1 Min. :821476 Min. :1172359 Min. :3.270e+02
## 1st Qu.:11558 1st Qu.:833029 1st Qu.:1181443 1st Qu.:1.482e+06
## Median :23149 Median :834305 Median :1183204 Median :1.494e+06
## Mean :23110 Mean :834102 Mean :1183264 Mean :2.175e+16
## 3rd Qu.:34664 3rd Qu.:835387 3rd Qu.:1185476 3rd Qu.:1.506e+06
## Max. :46172 Max. :842417 Max. :1193451 Max. :5.002e+20
##
## FECHA HORA DIA
## 2015-08-18T00:00:00.000Z: 183 07:00 AM: 951 Min. : 1.00
## 2015-08-28T00:00:00.000Z: 183 08:00 AM: 921 1st Qu.: 8.00
## 2015-09-11T00:00:00.000Z: 162 10:00 AM: 801 Median :16.00
## 2015-12-07T00:00:00.000Z: 161 07:30 AM: 767 Mean :15.74
## 2015-09-17T00:00:00.000Z: 160 06:30 AM: 756 3rd Qu.:23.00
## 2015-09-18T00:00:00.000Z: 160 06:00 AM: 741 Max. :31.00
## (Other) :41071 (Other) :37143
## PERIODO CLASE DIRECCION
## Min. :2015 Choque :28249 CR 64 C CL 78: 211
## 1st Qu.:2015 Atropello : 4485 CR 63 CL 44 : 177
## Median :2015 Otro : 4218 CR 57 CL 44 : 172
## Mean :2015 Caida Ocupante : 3673 CR 64 C CL 67: 132
## 3rd Qu.:2015 Volcamiento : 1435 CR 80 CL 50 : 122
## Max. :2015 Caída de Ocupante: 18 CR 52 CL 10 : 121
## (Other) : 2 (Other) :41145
## DIRECCION_ENC CBML
## CR 064 C 078 000 00000: 227 1019 : 1067
## CR 063 044 000 00000 : 177 0517 : 899
## CR 057 044 000 00000 : 173 1007 : 882
## CR 064 C 067 000 00000: 132 1006 : 867
## CR 080 050 000 00000 : 122 1012 : 804
## CR 052 010 000 00000 : 121 1105 : 783
## (Other) :41128 (Other):36778
## TIPO_GEOCOD GRAVEDAD
## Malla vial :31011 HERIDO :23273
## Malla vial cruce invertido : 2341 MUERTO : 250
## EPM sin Interior : 886 SOLO DAÑOS:18557
## EPM con Interior : 459
## Malla vial aproximada: CR 1-2 : 86
## Malla vial aproximada: CR 65-80: 85
## (Other) : 7212
## BARRIO COMUNA DISENO
## La Candelaria : 1069 La Candelaria : 9355 Tramo de via :33729
## Caribe : 899 Laureles Estadio: 4908 Interseccion : 5794
## Guayaquil : 883 Castilla : 4000 Lote o Predio: 1219
## San Benito : 868 El Poblado : 3194 Glorieta : 751
## Perpetuo Socorro : 804 Robledo : 2884 : 250
## Los Conquistadores: 785 Guayabal : 2814 Paso Elevado : 101
## (Other) :36772 (Other) :14925 (Other) : 236
## DIA_NOMBRE MES
## DOMINGO :4048 Min. : 1.000
## JUEVES :6516 1st Qu.: 4.000
## LUNES :5973 Median : 7.000
## MARTES :6696 Mean : 6.624
## MIÉRCOLES:6341 3rd Qu.:10.000
## SÁBADO :6003 Max. :12.000
## VIERNES :6503str(raw_data_2015)## 'data.frame': 42080 obs. of 19 variables:
## $ X.U.FEFF.OBJECTID: int 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 ...
## $ X : num 835767 839497 832971 831318 833711 ...
## $ Y : num 1183119 1177763 1178867 1187204 1185631 ...
## $ RADICADO : num 1471803 1471804 1471805 1471806 1471807 ...
## $ FECHA : Factor w/ 365 levels "2015-01-01T00:00:00.000Z",..: 13 13 13 13 13 12 13 13 13 13 ...
## $ HORA : Factor w/ 898 levels "01:00 AM","01:00 PM",..: 103 610 691 413 342 651 560 493 342 868 ...
## $ DIA : int 13 13 13 13 13 12 13 13 13 13 ...
## $ PERIODO : int 2015 2015 2015 2015 2015 2015 2015 2015 2015 2015 ...
## $ CLASE : Factor w/ 8 levels "","Atropello",..: 8 7 8 7 7 8 7 7 7 4 ...
## $ DIRECCION : Factor w/ 12729 levels "CL 1 A CR 35",..: 7461 5630 9229 4297 10332 5630 9640 10693 9766 9166 ...
## $ DIRECCION_ENC : Factor w/ 12523 levels "000","CL 010 000 00000",..: 7057 5546 8947 3700 9999 5546 9515 10334 9266 8875 ...
## $ CBML : Factor w/ 394 levels "","0","0101",..: 217 389 317 150 106 389 85 133 232 200 ...
## $ TIPO_GEOCOD : Factor w/ 2229 levels "Catastro con Interior",..: 5 1014 1645 797 1844 1014 1728 1892 5 5 ...
## $ GRAVEDAD : Factor w/ 3 levels "HERIDO","MUERTO",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ BARRIO : Factor w/ 319 levels "","0","6001",..: 51 283 62 210 99 283 170 76 194 248 ...
## $ COMUNA : Factor w/ 26 levels "","0","Aranjuez",..: 18 12 15 22 7 12 7 22 19 18 ...
## $ DISENO : Factor w/ 13 levels "","Ciclo Ruta",..: 11 11 11 11 11 11 11 11 11 11 ...
## $ DIA_NOMBRE : Factor w/ 7 levels "DOMINGO ","JUEVES ",..: 4 4 4 4 4 3 4 4 4 4 ...
## $ MES : int 1 1 1 1 1 1 1 1 1 1 ...levels(raw_data_2015$CLASE)## [1] "" "Atropello" "Caída de Ocupante"
## [4] "Caida Ocupante" "Choque" "Incendio"
## [7] "Otro" "Volcamiento"levels(raw_data_2015$CLASE) <- c("Otro","Atropello","Caida_Ocupante","Caida_Ocupante","Choque",
"Otro","Otro","Volcamiento")levels(raw_data_2015$CLASE)## [1] "Otro" "Atropello" "Caida_Ocupante" "Choque"
## [5] "Volcamiento"Datos 2016
raw_data_2016 <- read.csv(file="./data/Accidentalidad_georreferenciada_2016.csv", encoding="UTF-8", header=TRUE, sep=",")head(raw_data_2016)## X.U.FEFF.OBJECTID X Y RADICADO FECHA
## 1 259034 827277.9 1175177 1519832 2016-01-20T00:00:00.000Z
## 2 259035 835105.9 1183570 1519765 2016-01-20T00:00:00.000Z
## 3 259036 834811.1 1182025 1519752 2016-01-20T00:00:00.000Z
## 4 259037 836020.8 1178803 1519682 2016-01-20T00:00:00.000Z
## 5 259038 832264.6 1185704 1519685 2016-01-20T00:00:00.000Z
## 6 259040 831843.3 1183130 1519883 2016-01-20T00:00:00.000Z
## HORA DIA PERIODO CLASE DIRECCION
## 1 11:00 AM 20 2016 Choque CR 55 CL 48 A Sur
## 2 01:25 PM 20 2016 Choque CL 54 CR 51
## 3 02:35 PM 20 2016 Choque CR 46 CL 40
## 4 08:40 AM 20 2016 Choque CR 30 CL 10 C
## 5 10:30 AM 20 2016 Volcamiento CR 80 CL 64
## 6 10:45 AM 20 2016 Choque CR 78 CL 44
## DIRECCION_ENC CBML TIPO_GEOCOD
## 1 CR 055 S 048 A 000 00000 8000 Malla vial
## 2 CL 054 051 000 00000 1005 Malla vial
## 3 CR 046 040 000 00000 1013 Malla vial
## 4 CR 030 010 C 000 00000 1407 Malla vial
## 5 CR 080 064 000 00000 0705 Malla vial aproximada: CR 80-65
## 6 CR 078 044 000 00000 1112 Malla vial
## GRAVEDAD BARRIO
## 1 SOLO DAÑOS Cabecera San Antonio de Prado
## 2 SOLO DAÑOS Estación Villa
## 3 SOLO DAÑOS Barrio Colón
## 4 SOLO DAÑOS Las Lomas No.2
## 5 SOLO DAÑOS Facultad de Minas U. Nacional
## 6 SOLO DAÑOS El Velódromo
## COMUNA DISENO DIA_NOMBRE MES
## 1 Corregimiento de San Antonio de Prado Lote o Predio MIÉRCOLES 1
## 2 La Candelaria Tramo de via MIÉRCOLES 1
## 3 La Candelaria Tramo de via MIÉRCOLES 1
## 4 El Poblado Tramo de via MIÉRCOLES 1
## 5 Robledo Tramo de via MIÉRCOLES 1
## 6 Laureles Estadio Interseccion MIÉRCOLES 1str(raw_data_2016)## 'data.frame': 42841 obs. of 19 variables:
## $ X.U.FEFF.OBJECTID: int 259034 259035 259036 259037 259038 259040 259041 259042 259044 259045 ...
## $ X : num 827278 835106 834811 836021 832265 ...
## $ Y : num 1175177 1183570 1182025 1178803 1185704 ...
## $ RADICADO : num 1519832 1519765 1519752 1519682 1519685 ...
## $ FECHA : Factor w/ 366 levels "2016-01-01T00:00:00.000Z",..: 20 20 20 20 20 20 20 20 20 20 ...
## $ HORA : Factor w/ 1092 levels "01:00 AM","01:00 PM",..: 918 36 128 696 878 896 778 81 42 452 ...
## $ DIA : int 20 20 20 20 20 20 20 20 20 20 ...
## $ PERIODO : int 2016 2016 2016 2016 2016 2016 2016 2016 2016 2016 ...
## $ CLASE : Factor w/ 8 levels "","Atropello",..: 5 5 5 5 8 5 5 5 5 5 ...
## $ DIRECCION : Factor w/ 12962 levels " CR 59 CL 64 -32",..: 9087 3363 7866 6408 11595 11298 1313 78 1793 5222 ...
## $ DIRECCION_ENC : Factor w/ 12631 levels "000","CL 053 000 00000",..: 8828 2700 7540 6119 11176 10918 759 284 1275 201 ...
## $ CBML : Factor w/ 396 levels "","0","0101",..: 382 197 210 298 122 246 332 298 207 310 ...
## $ TIPO_GEOCOD : Factor w/ 2254 levels "Catastro con Interior",..: 5 5 5 5 2013 5 5 2252 5 5 ...
## $ GRAVEDAD : Factor w/ 3 levels "HERIDO","MUERTO",..: 3 3 3 3 3 3 3 3 3 3 ...
## $ BARRIO : Factor w/ 313 levels "","0","6001",..: 56 111 33 175 112 109 242 175 60 97 ...
## $ COMUNA : Factor w/ 25 levels "","0","Aranjuez",..: 9 18 18 14 22 19 5 14 18 14 ...
## $ DISENO : Factor w/ 13 levels "","Ciclo Ruta",..: 5 11 11 11 11 4 11 4 4 11 ...
## $ DIA_NOMBRE : Factor w/ 7 levels "DOMINGO ","JUEVES ",..: 5 5 5 5 5 5 5 5 5 5 ...
## $ MES : int 1 1 1 1 1 1 1 1 1 1 ...levels(raw_data_2016$CLASE)## [1] "" "Atropello" "Caída de Ocupante"
## [4] "Caida Ocupante" "Choque" "Incendio"
## [7] "Otro" "Volcamiento"levels(raw_data_2016$CLASE) <- c("Otro","Atropello","Caida_Ocupante","Caida_Ocupante","Choque",
"Otro","Otro","Volcamiento")levels(raw_data_2016$CLASE)## [1] "Otro" "Atropello" "Caida_Ocupante" "Choque"
## [5] "Volcamiento"Datos 2017
raw_data_2017 <- read.csv(file="./data/Accidentalidad_georreferenciada_2017.csv", encoding="UTF-8", header=TRUE, sep=",")head(raw_data_2017)## X.U.FEFF.OBJECTID X Y RADICADO FECHA
## 1 504352 834796.6 1181345 1590356 2017-07-20T00:00:00.000Z
## 2 504353 831461.0 1181883 1586285 2017-06-15T00:00:00.000Z
## 3 504354 833229.3 1186996 1588185 2017-07-02T00:00:00.000Z
## 4 504355 833366.8 1181955 1576853 2017-03-29T00:00:00.000Z
## 5 504356 832577.6 1184274 1591283 2017-07-26T00:00:00.000Z
## 6 504358 834246.7 1180257 1578106 2017-04-12T00:00:00.000Z
## HORA DIA PERIODO CLASE DIRECCION
## 1 08:00 AM 20 2017 Caida Ocupante CR 43 A CL 33
## 2 04:50 PM 15 2017 Choque CR 80 CL 33
## 3 04:20 PM 2 2017 Choque CR 80 CL 80 A
## 4 06:29 PM 29 2017 Choque CL 33 Norte CR 65
## 5 08:10 PM 26 2017 Atropello CL 50 CR 74
## 6 11:10 AM 12 2017 Choque CR 48 CL 20
## DIRECCION_ENC CBML TIPO_GEOCOD GRAVEDAD
## 1 CR 043 A 033 000 00000 1020 Malla vial HERIDO
## 2 CR 080 033 000 00000 1109 Malla vial SOLO DAÑOS
## 3 CR 080 080 A 000 00000 0710 Malla vial HERIDO
## 4 CL 033 065 000 00000 1105 Malla vial SOLO DAÑOS
## 5 CL 050 074 000 00000 1115 Malla vial HERIDO
## 6 CR 048 020 000 00000 1403 Malla vial SOLO DAÑOS
## BARRIO COMUNA DISENO DIA_NOMBRE MES
## 1 San Diego La Candelaria Interseccion JUEVES 7
## 2 Las Acacias Laureles Estadio Glorieta JUEVES 6
## 3 López de Mesa Robledo Tramo de via DOMINGO 7
## 4 Los Conquistadores Laureles Estadio Tramo de via MIÉRCOLES 3
## 5 Cuarta Brigada Laureles Estadio Tramo de via MIÉRCOLES 7
## 6 Villa Carlota El Poblado Tramo de via MIÉRCOLES 4str(raw_data_2017)## 'data.frame': 42563 obs. of 19 variables:
## $ X.U.FEFF.OBJECTID: int 504352 504353 504354 504355 504356 504358 504359 504360 504361 504362 ...
## $ X : num 834797 831461 833229 833367 832578 ...
## $ Y : num 1181345 1181883 1186996 1181955 1184274 ...
## $ RADICADO : num 1590356 1586285 1588185 1576853 1591283 ...
## $ FECHA : Factor w/ 370 levels "2017-01-01T00:00:00.000Z",..: 201 166 183 88 207 102 154 110 180 174 ...
## $ HORA : Factor w/ 1106 levels "01:00 AM","01:00 PM",..: 638 321 278 473 657 950 291 362 203 555 ...
## $ DIA : int 20 15 2 29 26 12 3 20 29 23 ...
## $ PERIODO : int 2017 2017 2017 2017 2017 2017 2017 2017 2017 2017 ...
## $ CLASE : Factor w/ 9 levels "Atropello","Caida Ocupante",..: 2 4 4 4 1 4 2 4 4 4 ...
## $ DIRECCION : Factor w/ 12879 levels "CL 1 A CR 70",..: 7319 11494 11596 1612 3122 7985 5410 3656 2392 652 ...
## $ DIRECCION_ENC : Factor w/ 12595 levels "000","CL 018 000 00000",..: 7116 11091 11186 1035 2406 7639 4443 2902 1738 5432 ...
## $ CBML : Factor w/ 403 levels "","0101","0102",..: 217 230 131 224 239 283 112 212 249 303 ...
## $ TIPO_GEOCOD : Factor w/ 2569 levels "Catastro con Interior",..: 5 5 5 5 5 5 5 5 5 4 ...
## $ GRAVEDAD : Factor w/ 3 levels "HERIDO","MUERTO",..: 1 3 1 3 1 3 1 1 3 3 ...
## $ BARRIO : Factor w/ 325 levels "","6001","9086",..: 252 169 185 195 78 314 134 322 89 135 ...
## $ COMUNA : Factor w/ 83 levels "","Alejandro Echavarría",..: 46 55 68 55 55 33 30 46 45 33 ...
## $ DISENO : Factor w/ 13 levels "","Ciclo Ruta",..: 4 3 11 11 11 11 5 11 4 4 ...
## $ DIA_NOMBRE : Factor w/ 7 levels "DOMINGO ","JUEVES ",..: 2 2 1 5 5 5 6 2 2 7 ...
## $ MES : int 7 6 7 3 7 4 6 4 6 6 ...levels(raw_data_2017$CLASE)## [1] "Atropello" "Caida Ocupante" "Caída Ocupante"
## [4] "Choque" "Choque " "Choque y Atropello"
## [7] "Incendio" "Otro" "Volcamiento"levels(raw_data_2017$CLASE) <- c("Atropello","Caida_Ocupante","Caida_Ocupante","Choque","Choque",
"Choque","Otro","Otro","Volcamiento")levels(raw_data_2017$CLASE)## [1] "Atropello" "Caida_Ocupante" "Choque" "Otro"
## [5] "Volcamiento"summary(raw_data_2014$CLASE)## Atropello Caida_Ocupante Choque Otro Volcamiento
## 4779 4157 27157 4529 972summary(raw_data_2015$CLASE)## Otro Atropello Caida_Ocupante Choque Volcamiento
## 4220 4485 3691 28249 1435summary(raw_data_2016$CLASE)## Otro Atropello Caida_Ocupante Choque Volcamiento
## 4879 4167 3680 28631 1484summary(raw_data_2017$CLASE)## Atropello Caida_Ocupante Choque Otro Volcamiento
## 3640 3433 29196 4722 1572raw_data_2018 <- read.csv(file="./data/Accidentalidad_georreferenciada_2018.csv", encoding="UTF-8", header=TRUE, sep=",")str(raw_data_2018)## 'data.frame': 40348 obs. of 19 variables:
## $ X.U.FEFF.OBJECTID: int 550556 550557 550558 550559 550560 550562 550563 550564 550565 550566 ...
## $ X : num 833196 833455 835882 831732 835720 ...
## $ Y : num 1184350 1187884 1183347 1180062 1181651 ...
## $ RADICADO : int 1612819 1612866 1612809 1612812 1612817 1612856 1612823 1612825 1612850 1612815 ...
## $ FECHA : Factor w/ 365 levels "2018-01-01T00:00:00.000Z",..: 14 14 14 14 14 14 14 14 14 14 ...
## $ HORA : Factor w/ 1485 levels "01:00 AM","01:00 PM",..: 57 57 168 168 182 222 281 334 348 372 ...
## $ DIA : int 14 14 14 14 14 14 14 14 14 14 ...
## $ PERIODO : int 2018 2018 2018 2018 2018 2018 2018 2018 2018 2018 ...
## $ CLASE : Factor w/ 7 levels "Atropello","Caida Ocupante",..: 4 4 2 6 4 1 4 4 4 2 ...
## $ DIRECCION : Factor w/ 12504 levels "CL 1 A CR 25",..: 10404 11159 7325 10853 6620 7455 5569 1588 2214 10721 ...
## $ DIRECCION_ENC : Factor w/ 12364 levels "000","CL 001 017 000 00000",..: 10180 10919 6991 10609 6409 7256 5538 1097 1679 10470 ...
## $ CBML : Factor w/ 375 levels "","0101","0102",..: 207 103 202 307 172 40 214 194 231 104 ...
## $ TIPO_GEOCOD : Factor w/ 2405 levels "Catastro con Interior",..: 5 5 5 5 5 3 5 5 5 5 ...
## $ GRAVEDAD : Factor w/ 3 levels "HERIDO","MUERTO",..: 1 3 1 1 3 1 3 3 3 1 ...
## $ BARRIO : Factor w/ 321 levels "","0","6001",..: 69 262 192 251 108 205 48 237 200 137 ...
## $ COMUNA : Factor w/ 24 levels "","Aranjuez",..: 18 12 17 4 5 19 18 17 16 12 ...
## $ DISENO : Factor w/ 13 levels "","Ciclo Ruta",..: 4 11 11 4 11 11 4 5 11 4 ...
## $ DIA_NOMBRE : Factor w/ 7 levels "DOMINGO ","JUEVES ",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ MES : int 1 1 1 1 1 1 1 1 1 1 ...levels(raw_data_2018$CLASE)## [1] "Atropello" "Caida Ocupante" "Caída Ocupante" "Choque"
## [5] "Incendio" "Otro" "Volcamiento"levels(raw_data_2018$CLASE) <- c("Atropello","Caida_Ocupante","Caida_Ocupante","Choque","Otro","Otro", "Volcamiento")levels(raw_data_2018$CLASE)## [1] "Atropello" "Caida_Ocupante" "Choque" "Otro"
## [5] "Volcamiento"summary(raw_data_2018$CLASE)## Atropello Caida_Ocupante Choque Otro Volcamiento
## 3604 3617 28207 3746 1174Union de los datos
Total_Dataset <- rbind(raw_data_2014,raw_data_2015,raw_data_2016, raw_data_2017, raw_data_2018)str(Total_Dataset)## 'data.frame': 209426 obs. of 19 variables:
## $ X.U.FEFF.OBJECTID: int 211279 211280 211281 211282 211283 211284 211285 211286 211287 211288 ...
## $ X : num 831190 835014 837033 830649 833741 ...
## $ Y : num 1179690 1184224 1184750 1181383 1188644 ...
## $ RADICADO : num 1423828 1423839 1423840 1423849 1423890 ...
## $ FECHA : Factor w/ 1831 levels "2014-01-01T00:00:00.000Z",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ HORA : Factor w/ 2239 levels "01:00 AM","01:00 PM",..: 95 954 1 938 777 201 469 268 37 72 ...
## $ DIA : int 1 1 1 1 1 1 1 1 1 1 ...
## $ PERIODO : int 2014 2014 2014 2014 2014 2014 2014 2014 2014 2014 ...
## $ CLASE : Factor w/ 5 levels "Atropello","Caida_Ocupante",..: 3 1 1 1 2 3 3 3 3 2 ...
## $ DIRECCION : Factor w/ 29182 levels "CL 1 A CR 42",..: 11441 8871 6987 1424 11297 8063 3081 6269 11142 8131 ...
## $ DIRECCION_ENC : Factor w/ 27999 levels "000","CL 029 000 00000",..: 11024 8642 6786 915 11166 7817 2469 6072 10863 7898 ...
## $ CBML : Factor w/ 699 levels "","0101","01010480009",..: 333 192 39 340 105 48 170 176 241 280 ...
## $ TIPO_GEOCOD : Factor w/ 5624 levels "Catastro con Interior",..: 2064 5 5 5 5 5 5 5 5 5 ...
## $ GRAVEDAD : Factor w/ 3 levels "HERIDO","MUERTO",..: 1 1 1 1 1 1 1 1 1 1 ...
## $ BARRIO : Factor w/ 340 levels "","6001","Aguas Frias",..: 181 129 201 175 81 40 31 155 294 302 ...
## $ COMUNA : Factor w/ 85 levels "","Aranjuez",..: 4 16 18 4 11 2 5 5 17 12 ...
## $ DISENO : Factor w/ 13 levels "","Ciclo Ruta",..: 11 11 11 11 11 4 11 4 11 11 ...
## $ DIA_NOMBRE : Factor w/ 7 levels "DOMINGO ","JUEVES ",..: 5 5 5 5 5 5 5 5 5 5 ...
## $ MES : int 1 1 1 1 1 1 1 1 1 1 ...levels(Total_Dataset$CLASE)## [1] "Atropello" "Caida_Ocupante" "Choque" "Otro"
## [5] "Volcamiento"summary(Total_Dataset$CLASE)## Atropello Caida_Ocupante Choque Otro Volcamiento
## 20675 18578 141440 22096 6637Para ahorra memoria se eliminan los dataframes por ano
rm(raw_data_2014,raw_data_2015,raw_data_2016, raw_data_2017, raw_data_2018)La Fecha esta cargada como factor, se realiza una transformacion a formato de fecha (date)
Total_Dataset$FECHA <- as.Date(Total_Dataset$FECHA, format="%Y-%m-%d")
tail(Total_Dataset)## X.U.FEFF.OBJECTID X Y RADICADO FECHA
## 209421 686550 836789.8 1186516 1652251 2018-12-05
## 209422 686551 834601.3 1186742 1652236 2018-12-05
## 209423 686552 832577.7 1183055 1652193 2018-12-05
## 209424 686553 833974.2 1183603 1652324 2018-12-05
## 209425 686554 834299.9 1180877 1652349 2018-12-05
## 209426 686555 836838.6 1186467 1652347 2018-12-05
## HORA DIA PERIODO CLASE DIRECCION
## 209421 03:20:00 p. m. 5 2018 Choque CR 44 A CL 88
## 209422 03:20:00 p. m. 5 2018 Choque CR 65 CL 91 C
## 209423 03:20:00 p. m. 5 2018 Choque CL 44 CR 71
## 209424 03:30:00 p. m. 5 2018 Choque CR 63 CL 50
## 209425 03:40:00 p. m. 5 2018 Choque CL 27 CR 46
## 209426 04:10:00 p. m. 5 2018 Otro CR 44 CL 88
## DIRECCION_ENC CBML
## 209421 CR 044 A 088 000 00000 03020770009
## 209422 CR 065 091 C 000 00000 0513
## 209423 CL 044 071 000 00000 1117
## 209424 CR 063 050 000 00000 1101
## 209425 CL 027 046 000 00000 1401
## 209426 CR 044 088 000 00000 0302
## TIPO_GEOCOD GRAVEDAD BARRIO
## 209421 Nomenclatura con Interior HERIDO Las Granjas
## 209422 Malla vial aproximada: CR 65-91A HERIDO Francisco Antonio Zea
## 209423 Malla vial SOLO DAÑOS Florida Nueva
## 209424 Malla vial HERIDO Carlos E. Restrepo
## 209425 Malla vial SOLO DAÑOS Barrio Colombia
## 209426 Malla vial HERIDO Las Granjas
## COMUNA DISENO DIA_NOMBRE MES
## 209421 Manrique Interseccion MIÉRCOLES 12
## 209422 Castilla Interseccion MIÉRCOLES 12
## 209423 Laureles Estadio Tramo de via MIÉRCOLES 12
## 209424 Laureles Estadio Tramo de via MIÉRCOLES 12
## 209425 El Poblado Lote o Predio MIÉRCOLES 12
## 209426 Manrique Interseccion MIÉRCOLES 12Se convierten las variables dia y mes a factor
Total_Dataset$DIA <- as.factor(Total_Dataset$DIA)
Total_Dataset$MES <- as.factor(Total_Dataset$MES)tail(Total_Dataset)## X.U.FEFF.OBJECTID X Y RADICADO FECHA
## 209421 686550 836789.8 1186516 1652251 2018-12-05
## 209422 686551 834601.3 1186742 1652236 2018-12-05
## 209423 686552 832577.7 1183055 1652193 2018-12-05
## 209424 686553 833974.2 1183603 1652324 2018-12-05
## 209425 686554 834299.9 1180877 1652349 2018-12-05
## 209426 686555 836838.6 1186467 1652347 2018-12-05
## HORA DIA PERIODO CLASE DIRECCION
## 209421 03:20:00 p. m. 5 2018 Choque CR 44 A CL 88
## 209422 03:20:00 p. m. 5 2018 Choque CR 65 CL 91 C
## 209423 03:20:00 p. m. 5 2018 Choque CL 44 CR 71
## 209424 03:30:00 p. m. 5 2018 Choque CR 63 CL 50
## 209425 03:40:00 p. m. 5 2018 Choque CL 27 CR 46
## 209426 04:10:00 p. m. 5 2018 Otro CR 44 CL 88
## DIRECCION_ENC CBML
## 209421 CR 044 A 088 000 00000 03020770009
## 209422 CR 065 091 C 000 00000 0513
## 209423 CL 044 071 000 00000 1117
## 209424 CR 063 050 000 00000 1101
## 209425 CL 027 046 000 00000 1401
## 209426 CR 044 088 000 00000 0302
## TIPO_GEOCOD GRAVEDAD BARRIO
## 209421 Nomenclatura con Interior HERIDO Las Granjas
## 209422 Malla vial aproximada: CR 65-91A HERIDO Francisco Antonio Zea
## 209423 Malla vial SOLO DAÑOS Florida Nueva
## 209424 Malla vial HERIDO Carlos E. Restrepo
## 209425 Malla vial SOLO DAÑOS Barrio Colombia
## 209426 Malla vial HERIDO Las Granjas
## COMUNA DISENO DIA_NOMBRE MES
## 209421 Manrique Interseccion MIÉRCOLES 12
## 209422 Castilla Interseccion MIÉRCOLES 12
## 209423 Laureles Estadio Tramo de via MIÉRCOLES 12
## 209424 Laureles Estadio Tramo de via MIÉRCOLES 12
## 209425 El Poblado Lote o Predio MIÉRCOLES 12
## 209426 Manrique Interseccion MIÉRCOLES 12Se genera frecuencia que es la tabla a analizar
library(sqldf)## Loading required package: gsubfn## Loading required package: proto## Loading required package: RSQLiteTotal_Dataset_Freq <- sqldf("SELECT FECHA, CLASE, count(CLASE) AS FREQ, DIA_NOMBRE, MES, DIA
FROM Total_Dataset
GROUP BY FECHA, CLASE")se agrega la variable ano al datafrate
Total_Dataset_Freq$ANO <- as.factor(format(Total_Dataset_Freq$FECHA,'%Y'))Se agrega la variable semana al dataframe
Total_Dataset_Freq$SEMANA <-as.factor(format(Total_Dataset_Freq$FECHA,'%V'))tail(Total_Dataset_Freq)## FECHA CLASE FREQ DIA_NOMBRE MES DIA ANO SEMANA
## 9023 2018-12-30 Volcamiento 2 DOMINGO 12 30 2018 52
## 9024 2018-12-31 Atropello 8 LUNES 12 31 2018 01
## 9025 2018-12-31 Caida_Ocupante 6 LUNES 12 31 2018 01
## 9026 2018-12-31 Choque 50 LUNES 12 31 2018 01
## 9027 2018-12-31 Otro 10 LUNES 12 31 2018 01
## 9028 2018-12-31 Volcamiento 2 LUNES 12 31 2018 01Se depura la variable DIA_NOMBRE para eliminar los espacios, quitar tildes, dar un orden al factor
levels(Total_Dataset_Freq$DIA_NOMBRE)## [1] "DOMINGO " "JUEVES " "LUNES " "MARTES " "MIÉRCOLES" "SÁBADO "
## [7] "VIERNES "levels(Total_Dataset_Freq$DIA_NOMBRE) <- c("DOMINGO","JUEVES","LUNES","MARTES","MIERCOLES","SABADO","VIERNES") levels(Total_Dataset_Freq$DIA_NOMBRE)## [1] "DOMINGO" "JUEVES" "LUNES" "MARTES" "MIERCOLES" "SABADO"
## [7] "VIERNES"Total_Dataset_Freq$DIA_NOMBRE <- ordered(Total_Dataset_Freq$DIA_NOMBRE,c("LUNES", "MARTES", "MIERCOLES", "JUEVES", "VIERNES", "SABADO", "DOMINGO"))levels(Total_Dataset_Freq$DIA_NOMBRE)## [1] "LUNES" "MARTES" "MIERCOLES" "JUEVES" "VIERNES" "SABADO"
## [7] "DOMINGO"Ahora se agregan otras variables a que caracterizan fechas especiales. Se carga un CSV y se realiza un left join a la tabla Total_Dataset_Freq
Dias_Especiales <- read.csv(file="./data/Caracterizacion.csv", encoding="UTF-8", header=TRUE, sep=";",stringsAsFactors=FALSE)head(Dias_Especiales)## Fecha Lunes martes miercoles jueves viernes sabado domingo Enero
## 1 1-1-2014 0 0 1 0 0 0 0 1
## 2 2-1-2014 0 0 0 1 0 0 0 1
## 3 3-1-2014 0 0 0 0 1 0 0 1
## 4 4-1-2014 0 0 0 0 0 1 0 1
## 5 5-1-2014 0 0 0 0 0 0 1 1
## 6 6-1-2014 1 0 0 0 0 0 0 1
## Febrero Marzo Abril Mayo Junio Julio Agosto Septiembre Octubre Noviembre
## 1 0 0 0 0 0 0 0 0 0 0
## 2 0 0 0 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 0 0 0 0
## 4 0 0 0 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0 0 0 0
## Diciembre Feriado Semana.Santa Prima Mujer Padre Madre AmoryAmistad
## 1 0 1 0 0 0 0 0 0
## 2 0 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 0 0
## 4 0 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0 0
## 6 0 1 0 0 0 0 0 0
## Viernes_Antes_Puente Quincena Viernes_Desp_Qincena
## 1 0 0 0
## 2 0 0 0
## 3 1 0 0
## 4 0 0 0
## 5 0 0 0
## 6 0 0 0Se transforma la variable FEcha a formato Fecha pues se cargo como factor y se unifica el nombre a FECHA para que coincida con l clave de la tabla Total_Dataset_Freq
Dias_Especiales <- Dias_Especiales[1:1826,]
Dias_Especiales$FECHA <- as.Date(Dias_Especiales$Fecha, format="%d-%m-%Y")Dias_Especiales$Fecha <- NULLEstos dias especiales se incorporan al dataframe Total_Dataset_Freq
Total_Dataset_Freq <- sqldf("SELECT *
FROM Total_Dataset_Freq
LEFT JOIN Dias_Especiales USING(FECHA)")head(Total_Dataset_Freq)## FECHA CLASE FREQ DIA_NOMBRE MES DIA ANO SEMANA Lunes
## 1 2014-01-01 Atropello 13 MIERCOLES 1 1 2014 01 0
## 2 2014-01-01 Caida_Ocupante 7 MIERCOLES 1 1 2014 01 0
## 3 2014-01-01 Choque 35 MIERCOLES 1 1 2014 01 0
## 4 2014-01-01 Otro 18 MIERCOLES 1 1 2014 01 0
## 5 2014-01-01 Volcamiento 1 MIERCOLES 1 1 2014 01 0
## 6 2014-01-02 Atropello 12 JUEVES 1 2 2014 01 0
## martes miercoles jueves viernes sabado domingo Enero Febrero Marzo Abril
## 1 0 1 0 0 0 0 1 0 0 0
## 2 0 1 0 0 0 0 1 0 0 0
## 3 0 1 0 0 0 0 1 0 0 0
## 4 0 1 0 0 0 0 1 0 0 0
## 5 0 1 0 0 0 0 1 0 0 0
## 6 0 0 1 0 0 0 1 0 0 0
## Mayo Junio Julio Agosto Septiembre Octubre Noviembre Diciembre Feriado
## 1 0 0 0 0 0 0 0 0 1
## 2 0 0 0 0 0 0 0 0 1
## 3 0 0 0 0 0 0 0 0 1
## 4 0 0 0 0 0 0 0 0 1
## 5 0 0 0 0 0 0 0 0 1
## 6 0 0 0 0 0 0 0 0 0
## Semana.Santa Prima Mujer Padre Madre AmoryAmistad Viernes_Antes_Puente
## 1 0 0 0 0 0 0 0
## 2 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 0
## 4 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0
## Quincena Viernes_Desp_Qincena
## 1 0 0
## 2 0 0
## 3 0 0
## 4 0 0
## 5 0 0
## 6 0 0ANALISIS DESCRIPTIVO
Para el Total de Accidentes
library(plotly)## Loading required package: ggplot2##
## Attaching package: 'plotly'## The following object is masked from 'package:ggplot2':
##
## last_plot## The following object is masked from 'package:stats':
##
## filter## The following object is masked from 'package:graphics':
##
## layoutplot_ly(data=Total_Dataset_Freq,
x = ~FECHA,
y = ~FREQ,
type = "scatter", mode = "lines",
split = ~ANO,
ine=list(width=1))%>%
layout(title='Accidentes_Medellin',
xaxis=list(title="Dia"),
yaxis=list(title="Unidades"))## Warning: 'scatter' objects don't have these attributes: 'ine'
## Valid attributes include:
## 'type', 'visible', 'showlegend', 'legendgroup', 'opacity', 'name', 'uid', 'ids', 'customdata', 'selectedpoints', 'hoverinfo', 'hoverlabel', 'stream', 'transforms', 'uirevision', 'x', 'x0', 'dx', 'y', 'y0', 'dy', 'stackgroup', 'orientation', 'groupnorm', 'stackgaps', 'text', 'hovertext', 'mode', 'hoveron', 'hovertemplate', 'line', 'connectgaps', 'cliponaxis', 'fill', 'fillcolor', 'marker', 'selected', 'unselected', 'textposition', 'textfont', 'r', 't', 'error_x', 'error_y', 'xcalendar', 'ycalendar', 'xaxis', 'yaxis', 'idssrc', 'customdatasrc', 'hoverinfosrc', 'xsrc', 'ysrc', 'textsrc', 'hovertextsrc', 'hovertemplatesrc', 'textpositionsrc', 'rsrc', 'tsrc', 'key', 'set', 'frame', 'transforms', '_isNestedKey', '_isSimpleKey', '_isGraticule', '_bbox'
## Warning: 'scatter' objects don't have these attributes: 'ine'
## Valid attributes include:
## 'type', 'visible', 'showlegend', 'legendgroup', 'opacity', 'name', 'uid', 'ids', 'customdata', 'selectedpoints', 'hoverinfo', 'hoverlabel', 'stream', 'transforms', 'uirevision', 'x', 'x0', 'dx', 'y', 'y0', 'dy', 'stackgroup', 'orientation', 'groupnorm', 'stackgaps', 'text', 'hovertext', 'mode', 'hoveron', 'hovertemplate', 'line', 'connectgaps', 'cliponaxis', 'fill', 'fillcolor', 'marker', 'selected', 'unselected', 'textposition', 'textfont', 'r', 't', 'error_x', 'error_y', 'xcalendar', 'ycalendar', 'xaxis', 'yaxis', 'idssrc', 'customdatasrc', 'hoverinfosrc', 'xsrc', 'ysrc', 'textsrc', 'hovertextsrc', 'hovertemplatesrc', 'textpositionsrc', 'rsrc', 'tsrc', 'key', 'set', 'frame', 'transforms', '_isNestedKey', '_isSimpleKey', '_isGraticule', '_bbox'
## Warning: 'scatter' objects don't have these attributes: 'ine'
## Valid attributes include:
## 'type', 'visible', 'showlegend', 'legendgroup', 'opacity', 'name', 'uid', 'ids', 'customdata', 'selectedpoints', 'hoverinfo', 'hoverlabel', 'stream', 'transforms', 'uirevision', 'x', 'x0', 'dx', 'y', 'y0', 'dy', 'stackgroup', 'orientation', 'groupnorm', 'stackgaps', 'text', 'hovertext', 'mode', 'hoveron', 'hovertemplate', 'line', 'connectgaps', 'cliponaxis', 'fill', 'fillcolor', 'marker', 'selected', 'unselected', 'textposition', 'textfont', 'r', 't', 'error_x', 'error_y', 'xcalendar', 'ycalendar', 'xaxis', 'yaxis', 'idssrc', 'customdatasrc', 'hoverinfosrc', 'xsrc', 'ysrc', 'textsrc', 'hovertextsrc', 'hovertemplatesrc', 'textpositionsrc', 'rsrc', 'tsrc', 'key', 'set', 'frame', 'transforms', '_isNestedKey', '_isSimpleKey', '_isGraticule', '_bbox'
## Warning: 'scatter' objects don't have these attributes: 'ine'
## Valid attributes include:
## 'type', 'visible', 'showlegend', 'legendgroup', 'opacity', 'name', 'uid', 'ids', 'customdata', 'selectedpoints', 'hoverinfo', 'hoverlabel', 'stream', 'transforms', 'uirevision', 'x', 'x0', 'dx', 'y', 'y0', 'dy', 'stackgroup', 'orientation', 'groupnorm', 'stackgaps', 'text', 'hovertext', 'mode', 'hoveron', 'hovertemplate', 'line', 'connectgaps', 'cliponaxis', 'fill', 'fillcolor', 'marker', 'selected', 'unselected', 'textposition', 'textfont', 'r', 't', 'error_x', 'error_y', 'xcalendar', 'ycalendar', 'xaxis', 'yaxis', 'idssrc', 'customdatasrc', 'hoverinfosrc', 'xsrc', 'ysrc', 'textsrc', 'hovertextsrc', 'hovertemplatesrc', 'textpositionsrc', 'rsrc', 'tsrc', 'key', 'set', 'frame', 'transforms', '_isNestedKey', '_isSimpleKey', '_isGraticule', '_bbox'
## Warning: 'scatter' objects don't have these attributes: 'ine'
## Valid attributes include:
## 'type', 'visible', 'showlegend', 'legendgroup', 'opacity', 'name', 'uid', 'ids', 'customdata', 'selectedpoints', 'hoverinfo', 'hoverlabel', 'stream', 'transforms', 'uirevision', 'x', 'x0', 'dx', 'y', 'y0', 'dy', 'stackgroup', 'orientation', 'groupnorm', 'stackgaps', 'text', 'hovertext', 'mode', 'hoveron', 'hovertemplate', 'line', 'connectgaps', 'cliponaxis', 'fill', 'fillcolor', 'marker', 'selected', 'unselected', 'textposition', 'textfont', 'r', 't', 'error_x', 'error_y', 'xcalendar', 'ycalendar', 'xaxis', 'yaxis', 'idssrc', 'customdatasrc', 'hoverinfosrc', 'xsrc', 'ysrc', 'textsrc', 'hovertextsrc', 'hovertemplatesrc', 'textpositionsrc', 'rsrc', 'tsrc', 'key', 'set', 'frame', 'transforms', '_isNestedKey', '_isSimpleKey', '_isGraticule', '_bbox'plot_ly(data=Total_Dataset_Freq,
x = ~ANO,
y = ~FREQ,
type = "box")%>%
layout(title='Accidentes_Medellin',
xaxis=list(title="ano"),
yaxis=list(title="Unidades"))plot_ly(data=Total_Dataset_Freq,
x = ~MES,
y = ~FREQ,
type = "box")%>%
layout(title='Accidentes_Medellin',
xaxis=list(title="Mes"),
yaxis=list(title="Unidades"))plot_ly(data=Total_Dataset_Freq,
x = ~DIA_NOMBRE,
y = ~FREQ,
type = "box")%>%
layout(title='Accidentes_Medellin',
xaxis=list(title="Dia_Nombre"),
yaxis=list(title="Unidades"))Se observa que excepto el domingo que es menor, los demasla accidentalidad promedio es muy similar, pero se diferencia en la cola derecha de la distribucion
plot_ly(data=Total_Dataset_Freq,
x = ~SEMANA,
y = ~FREQ,
type = "box")%>%
layout(title='Accidentes_Medellin',
xaxis=list(title="SEMANA"),
yaxis=list(title="Unidades"))attach(Total_Dataset_Freq)aggregate(FREQ~ANO*SEMANA, data=Total_Dataset_Freq,FUN=mean)%>%
plot_ly(x = ~SEMANA,
y = ~FREQ,
type = "scatter" ,mode = "lines",
split = ~ANO,
line=list(width=1))%>%
layout(title='Promedio diario mensual de ACCIDENTES en Medellin',
xaxis=list(title="SEMANA"),
yaxis=list(title="No. Accidentes"))aggregate(FREQ~ANO*MES, data=Total_Dataset_Freq,FUN=mean)%>%
plot_ly(x = ~MES,
y = ~FREQ,
type = "scatter" ,mode = "lines",
split = ~ANO,
line=list(width=1))%>%
layout(title='Promedio diario mensual de ACCIDENTES en Medellin',
xaxis=list(title="Mes"),
yaxis=list(title="No. Accidentes"))aggregate(FREQ~CLASE*MES, data=Total_Dataset_Freq,FUN=mean)%>%
plot_ly(x = ~MES,
y = ~FREQ,
type = "scatter" ,mode = "lines",
split = ~CLASE,
line=list(width=1))%>%
layout(title='Promedio diario mensual de ACCIDENTES en Medellin por CLASE',
xaxis=list(title="Mes"),
yaxis=list(title="No. Accidentes"))aggregate(FREQ~CLASE*SEMANA, data=Total_Dataset_Freq,FUN=mean)%>%
plot_ly(x = ~SEMANA,
y = ~FREQ,
type = "scatter" ,mode = "lines",
split = ~CLASE,
line=list(width=1))%>%
layout(title='Promedio diario SEMANAL de ACCIDENTES en Medellin por CLASE',
xaxis=list(title="Mes"),
yaxis=list(title="No. Accidentes"))aggregate(FREQ~ANO*DIA_NOMBRE, data=Total_Dataset_Freq,FUN=mean)%>%
plot_ly(x = ~DIA_NOMBRE,
y = ~FREQ,
type = "scatter" ,mode = "lines",
split = ~ANO,
line=list(width=1))%>%
layout(title='Promedio diario DIA DE LA SEMANA de Accidentes de Transito en Medellin',
xaxis=list(title="Dia Semana"),
yaxis=list(title="No. Accidentes"))aggregate(FREQ~ANO*CLASE, data=Total_Dataset_Freq,FUN=mean)%>%
plot_ly(x = ~CLASE,
y = ~FREQ,
type = "scatter" ,mode = "lines",
split = ~ANO,
line=list(width=1))%>%
layout(title='Promedio diario CLASE de Accidentes de Transito en Medellin',
xaxis=list(title="Clase"),
yaxis=list(title="No. Accidentes"))LAs predicciones a realizar seran para cada uno de las 5 clases de accidentes, y para los periodos de tiempo mensual, semanal y diario, por lo que se generaran 15 modelos, 5 clases de accidente por 3 tipos de periodo de prediccion.
Se realiza entonces la particion de los datos por frecuencia: M: Mensual, S: Semanal, D: Diario
primero se realiza un reshape del dataframe para consolidar los casos por dia y separarlos factores CLASE en columnas
library(reshape)##
## Attaching package: 'reshape'## The following object is masked from 'package:plotly':
##
## renameTotal_Dataset_Freq <- cast(Total_Dataset_Freq[,c(1,2,3)],FECHA~CLASE)## Using FREQ as value column. Use the value argument to cast to override this choiceTotal_Dataset_Freq <- sqldf("SELECT *
FROM Total_Dataset_Freq
LEFT JOIN Dias_Especiales USING(FECHA)")se agrega la variable ano al datafrate
Total_Dataset_Freq$ANO <- as.factor(format(Total_Dataset_Freq$FECHA,'%Y'))Se agrega la variable semana al dataframe
Total_Dataset_Freq$SEMANA <-as.factor(format(Total_Dataset_Freq$FECHA,'%V'))SE agrega la variable dia al dataframe
Total_Dataset_Freq$DIA <-as.factor(format(Total_Dataset_Freq$FECHA,'%d'))Se agrega el dia de la semana
Total_Dataset_Freq$DIA_SEMANA <-as.factor(weekdays(Total_Dataset_Freq$FECHA))Se agrega el mes
Total_Dataset_Freq$MES <-as.factor(format(Total_Dataset_Freq$FECHA,'%m'))head(Total_Dataset_Freq)## FECHA Atropello Caida_Ocupante Choque Otro Volcamiento Lunes martes
## 1 2014-01-01 13 7 35 18 1 0 0
## 2 2014-01-02 12 7 43 9 1 0 0
## 3 2014-01-03 7 5 67 13 1 0 0
## 4 2014-01-04 11 7 40 9 1 0 0
## 5 2014-01-05 6 5 43 10 3 0 0
## 6 2014-01-06 4 5 23 9 2 1 0
## miercoles jueves viernes sabado domingo Enero Febrero Marzo Abril Mayo
## 1 1 0 0 0 0 1 0 0 0 0
## 2 0 1 0 0 0 1 0 0 0 0
## 3 0 0 1 0 0 1 0 0 0 0
## 4 0 0 0 1 0 1 0 0 0 0
## 5 0 0 0 0 1 1 0 0 0 0
## 6 0 0 0 0 0 1 0 0 0 0
## Junio Julio Agosto Septiembre Octubre Noviembre Diciembre Feriado
## 1 0 0 0 0 0 0 0 1
## 2 0 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 0 0
## 4 0 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0 1
## Semana.Santa Prima Mujer Padre Madre AmoryAmistad Viernes_Antes_Puente
## 1 0 0 0 0 0 0 0
## 2 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 1
## 4 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0
## Quincena Viernes_Desp_Qincena ANO SEMANA DIA DIA_SEMANA MES
## 1 0 0 2014 01 01 miércoles 01
## 2 0 0 2014 01 02 jueves 01
## 3 0 0 2014 01 03 viernes 01
## 4 0 0 2014 01 04 sábado 01
## 5 0 0 2014 01 05 domingo 01
## 6 0 0 2014 02 06 lunes 01Se reemplazan los NA por ceros
Total_Dataset_Freq[is.na(Total_Dataset_Freq)] <- 0Se agrega la columna Total_Accidentes
Total_Dataset_Freq$Total_Accidentes <- Total_Dataset_Freq$Atropello + Total_Dataset_Freq$Caida_Ocupante + Total_Dataset_Freq$Choque + Total_Dataset_Freq$Otro + Total_Dataset_Freq$VolcamientoSe agrega la columna Accidentes Graves
Total_Dataset_Freq$Accidentes_Graves <- Total_Dataset_Freq$Atropello + Total_Dataset_Freq$Caida_Ocupante + Total_Dataset_Freq$Volcamientotail(Total_Dataset_Freq)## FECHA Atropello Caida_Ocupante Choque Otro Volcamiento Lunes
## 1821 2018-12-26 11 7 84 6 4 0
## 1822 2018-12-27 13 3 77 9 2 0
## 1823 2018-12-28 6 6 84 7 1 0
## 1824 2018-12-29 15 4 59 7 3 0
## 1825 2018-12-30 5 7 33 10 2 0
## 1826 2018-12-31 8 6 50 10 2 1
## martes miercoles jueves viernes sabado domingo Enero Febrero Marzo
## 1821 0 1 0 0 0 0 0 0 0
## 1822 0 0 1 0 0 0 0 0 0
## 1823 0 0 0 1 0 0 0 0 0
## 1824 0 0 0 0 1 0 0 0 0
## 1825 0 0 0 0 0 1 0 0 0
## 1826 0 0 0 0 0 0 0 0 0
## Abril Mayo Junio Julio Agosto Septiembre Octubre Noviembre Diciembre
## 1821 0 0 0 0 0 0 0 0 1
## 1822 0 0 0 0 0 0 0 0 1
## 1823 0 0 0 0 0 0 0 0 1
## 1824 0 0 0 0 0 0 0 0 1
## 1825 0 0 0 0 0 0 0 0 1
## 1826 0 0 0 0 0 0 0 0 1
## Feriado Semana.Santa Prima Mujer Padre Madre AmoryAmistad
## 1821 0 0 0 0 0 0 0
## 1822 0 0 0 0 0 0 0
## 1823 0 0 0 0 0 0 0
## 1824 0 0 0 0 0 0 0
## 1825 0 0 0 0 0 0 0
## 1826 0 0 0 0 0 0 0
## Viernes_Antes_Puente Quincena Viernes_Desp_Qincena ANO SEMANA DIA
## 1821 0 0 0 2018 52 26
## 1822 0 0 0 2018 52 27
## 1823 0 0 0 2018 52 28
## 1824 0 0 0 2018 52 29
## 1825 0 0 0 2018 52 30
## 1826 0 1 0 2018 01 31
## DIA_SEMANA MES Total_Accidentes Accidentes_Graves
## 1821 miércoles 12 112 22
## 1822 jueves 12 104 18
## 1823 viernes 12 104 13
## 1824 sábado 12 88 22
## 1825 domingo 12 57 14
## 1826 lunes 12 76 16EVALUACION DE MODELOS
1. Particion de los datos para entrenamiento y test
Train_D_Dataset <- subset(Total_Dataset_Freq, ANO!="2018")summary(Train_D_Dataset$ANO)## 2014 2015 2016 2017 2018
## 365 365 366 365 0Se ajustan otra vez los niveles del factor ANO
Train_D_Dataset$ANO <- factor(Train_D_Dataset$ANO)summary(Train_D_Dataset$ANO)## 2014 2015 2016 2017
## 365 365 366 365library(sqldf)
Test_D_Dataset <- sqldf("SELECT *
FROM Total_Dataset_Freq
WHERE ANO == 2018")summary(Test_D_Dataset$ANO)## 2014 2015 2016 2017 2018
## 0 0 0 0 365Se ajustan otra vez los niveles del factor ANO
Test_D_Dataset$ANO <- factor(Test_D_Dataset$ANO)summary(Test_D_Dataset$ANO)## 2018
## 3652. KNN
Total Accidentes
set.seed(123) # fija la semilla del generador de números para que sea reproduciblehead(Train_D_Dataset)## FECHA Atropello Caida_Ocupante Choque Otro Volcamiento Lunes martes
## 1 2014-01-01 13 7 35 18 1 0 0
## 2 2014-01-02 12 7 43 9 1 0 0
## 3 2014-01-03 7 5 67 13 1 0 0
## 4 2014-01-04 11 7 40 9 1 0 0
## 5 2014-01-05 6 5 43 10 3 0 0
## 6 2014-01-06 4 5 23 9 2 1 0
## miercoles jueves viernes sabado domingo Enero Febrero Marzo Abril Mayo
## 1 1 0 0 0 0 1 0 0 0 0
## 2 0 1 0 0 0 1 0 0 0 0
## 3 0 0 1 0 0 1 0 0 0 0
## 4 0 0 0 1 0 1 0 0 0 0
## 5 0 0 0 0 1 1 0 0 0 0
## 6 0 0 0 0 0 1 0 0 0 0
## Junio Julio Agosto Septiembre Octubre Noviembre Diciembre Feriado
## 1 0 0 0 0 0 0 0 1
## 2 0 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 0 0
## 4 0 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0 1
## Semana.Santa Prima Mujer Padre Madre AmoryAmistad Viernes_Antes_Puente
## 1 0 0 0 0 0 0 0
## 2 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 1
## 4 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0
## Quincena Viernes_Desp_Qincena ANO SEMANA DIA DIA_SEMANA MES
## 1 0 0 2014 01 01 miércoles 01
## 2 0 0 2014 01 02 jueves 01
## 3 0 0 2014 01 03 viernes 01
## 4 0 0 2014 01 04 sábado 01
## 5 0 0 2014 01 05 domingo 01
## 6 0 0 2014 02 06 lunes 01
## Total_Accidentes Accidentes_Graves
## 1 74 21
## 2 72 20
## 3 93 13
## 4 68 19
## 5 67 14
## 6 43 11#DIA_NOMBRE+DIA+SEMANA+MES+ANO+AmoryAmistad+Madre+Padre+Mujer+Prima+Semana.Santa+Feriado
library(caret)## Loading required package: latticetrcntrl = trainControl(method="cv", number=10)
caret_knn_fit = caret::train(Total_Accidentes~SEMANA+DIA_SEMANA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente, data=Train_D_Dataset,
method = "knn", trControl = trcntrl,
preProcess=c("center", "scale"),
tuneLength = 10)summary(caret_knn_fit)## Length Class Mode
## learn 2 -none- list
## k 1 -none- numeric
## theDots 0 -none- list
## xNames 63 -none- character
## problemType 1 -none- character
## tuneValue 1 data.frame list
## obsLevels 1 -none- logical
## param 0 -none- listcaret_knn_fit## k-Nearest Neighbors
##
## 1461 samples
## 7 predictor
##
## Pre-processing: centered (63), scaled (63)
## Resampling: Cross-Validated (10 fold)
## Summary of sample sizes: 1313, 1315, 1316, 1316, 1315, 1315, ...
## Resampling results across tuning parameters:
##
## k RMSE Rsquared MAE
## 5 30.54405 0.04852764 26.07710
## 7 27.82936 0.05057528 23.31746
## 9 26.25975 0.06452191 21.62607
## 11 25.54304 0.07395262 20.70324
## 13 25.25037 0.07830750 20.13094
## 15 25.13014 0.07935002 19.86330
## 17 25.13203 0.07863251 19.73166
## 19 25.08799 0.08092453 19.64728
## 21 25.16812 0.07471798 19.70308
## 23 25.05694 0.08113243 19.88569
##
## RMSE was used to select the optimal model using the smallest value.
## The final value used for the model was k = 23.Calculo MSE y RMSE para los datos de entrenamiento
y_tr_pred_knn<-predict(caret_knn_fit,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_tr_knn<-mean((Train_D_Dataset$Total_Accidentes-y_tr_pred_knn)^2) # calcula el mse de entrenamiento
RMSE_tr_knn = sqrt(mse_tr_knn)
mse_tr_knn## [1] 576.2214RMSE_tr_knn## [1] 24.00461Calculo MSE y RMSE para los datos de prueba (2018)
y_test_pred_knn<-predict(caret_knn_fit,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_test_knn<-mean((Test_D_Dataset$Total_Accidentes-y_test_pred_knn)^2) # calcula el mse de entrenamiento
RMSE_test_knn = sqrt(mse_test_knn)
mse_test_knn## [1] 629.9186RMSE_test_knn## [1] 25.09818plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_knn,
name='Modelo knn',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))Choques
trcntrl = trainControl(method="cv", number=10)
caret_knn_fit_Ch = caret::train(Choque~SEMANA+DIA_SEMANA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente, data=Train_D_Dataset,
method = "knn", trControl = trcntrl,
preProcess=c("center", "scale"),
tuneLength = 10)summary(caret_knn_fit_Ch)## Length Class Mode
## learn 2 -none- list
## k 1 -none- numeric
## theDots 0 -none- list
## xNames 63 -none- character
## problemType 1 -none- character
## tuneValue 1 data.frame list
## obsLevels 1 -none- logical
## param 0 -none- listcaret_knn_fit_Ch## k-Nearest Neighbors
##
## 1461 samples
## 7 predictor
##
## Pre-processing: centered (63), scaled (63)
## Resampling: Cross-Validated (10 fold)
## Summary of sample sizes: 1315, 1314, 1317, 1314, 1315, 1314, ...
## Resampling results across tuning parameters:
##
## k RMSE Rsquared MAE
## 5 24.74517 0.04171384 21.39042
## 7 22.53987 0.04289098 19.05299
## 9 21.28711 0.05153353 17.68713
## 11 20.58471 0.06268419 16.70616
## 13 20.29705 0.06945246 16.32945
## 15 20.15232 0.07633053 16.04545
## 17 20.14957 0.07621763 15.94365
## 19 20.09285 0.08181138 15.86564
## 21 20.04423 0.08383903 15.85666
## 23 20.14198 0.07589901 16.05049
##
## RMSE was used to select the optimal model using the smallest value.
## The final value used for the model was k = 21.y_tr_pred_knn_ch<-predict(caret_knn_fit_Ch,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_tr_knn_ch<-mean((Train_D_Dataset$Choque-y_tr_pred_knn_ch)^2) # calcula el mse de entrenamiento
RMSE_tr_knn_ch = sqrt(mse_tr_knn_ch)
mse_tr_knn_ch## [1] 370.3256RMSE_tr_knn_ch## [1] 19.24385y_test_pred_knn_ch<-predict(caret_knn_fit_Ch,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_test_knn_ch<-mean((Test_D_Dataset$Choque-y_test_pred_knn_ch)^2) # calcula el mse de entrenamiento
RMSE_test_knn_ch = sqrt(mse_test_knn_ch)
mse_test_knn_ch## [1] 380.936RMSE_test_knn_ch## [1] 19.51758plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Choque,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_knn_ch,
name='Modelo knn',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Choques',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))Accidentes Graves
trcntrl = trainControl(method="cv", number=10)
caret_knn_fit_gr = caret::train(Accidentes_Graves~SEMANA+DIA_SEMANA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente, data=Train_D_Dataset,
method = "knn", trControl = trcntrl,
preProcess=c("center", "scale"),
tuneLength = 10)summary(caret_knn_fit_gr)## Length Class Mode
## learn 2 -none- list
## k 1 -none- numeric
## theDots 0 -none- list
## xNames 63 -none- character
## problemType 1 -none- character
## tuneValue 1 data.frame list
## obsLevels 1 -none- logical
## param 0 -none- listcaret_knn_fit_gr## k-Nearest Neighbors
##
## 1461 samples
## 7 predictor
##
## Pre-processing: centered (63), scaled (63)
## Resampling: Cross-Validated (10 fold)
## Summary of sample sizes: 1316, 1315, 1315, 1315, 1314, 1315, ...
## Resampling results across tuning parameters:
##
## k RMSE Rsquared MAE
## 5 7.170948 0.02435963 5.594836
## 7 6.876460 0.03325676 5.374907
## 9 6.773950 0.03372694 5.317876
## 11 6.706650 0.03861691 5.264893
## 13 6.661325 0.04345227 5.255031
## 15 6.624616 0.05005969 5.207789
## 17 6.642902 0.04657338 5.224121
## 19 6.638075 0.04545474 5.210832
## 21 6.643289 0.04463043 5.202793
## 23 6.640412 0.04467729 5.187826
##
## RMSE was used to select the optimal model using the smallest value.
## The final value used for the model was k = 15.y_tr_pred_knn_gr<-predict(caret_knn_fit_gr,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_tr_knn_gr<-mean((Train_D_Dataset$Accidentes_Graves-y_tr_pred_knn_gr)^2) # calcula el mse de entrenamiento
RMSE_tr_knn_gr = sqrt(mse_tr_knn_gr)
mse_tr_knn_gr## [1] 40.21077RMSE_tr_knn_gr## [1] 6.341196y_test_pred_knn_gr<-predict(caret_knn_fit_gr,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_test_knn_gr<-mean((Test_D_Dataset$Accidentes_Graves-y_test_pred_knn_gr)^2) # calcula el mse de entrenamiento
RMSE_test_knn_gr = sqrt(mse_test_knn_gr)
mse_test_knn_gr## [1] 45.44197RMSE_test_knn_gr## [1] 6.741066plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Accidentes_Graves,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_knn_gr,
name='Modelo knn',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Accidentes Graves',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))Otros
trcntrl = trainControl(method="cv", number=10)
caret_knn_fit_ot = caret::train(Otro~SEMANA+DIA_SEMANA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente, data=Train_D_Dataset,
method = "knn", trControl = trcntrl,
preProcess=c("center", "scale"),
tuneLength = 10)summary(caret_knn_fit_ot)## Length Class Mode
## learn 2 -none- list
## k 1 -none- numeric
## theDots 0 -none- list
## xNames 63 -none- character
## problemType 1 -none- character
## tuneValue 1 data.frame list
## obsLevels 1 -none- logical
## param 0 -none- listcaret_knn_fit_ot## k-Nearest Neighbors
##
## 1461 samples
## 7 predictor
##
## Pre-processing: centered (63), scaled (63)
## Resampling: Cross-Validated (10 fold)
## Summary of sample sizes: 1315, 1315, 1315, 1316, 1315, 1314, ...
## Resampling results across tuning parameters:
##
## k RMSE Rsquared MAE
## 5 5.208633 0.01213720 3.976676
## 7 5.075883 0.01690476 3.880784
## 9 5.039922 0.01559127 3.874442
## 11 4.997860 0.01704169 3.851977
## 13 4.970320 0.01533633 3.854434
## 15 4.966180 0.01488424 3.851711
## 17 4.955894 0.01507451 3.849701
## 19 4.942486 0.01743955 3.838876
## 21 4.927215 0.01903057 3.831270
## 23 4.914770 0.01931428 3.813755
##
## RMSE was used to select the optimal model using the smallest value.
## The final value used for the model was k = 23.y_tr_pred_knn_ot<-predict(caret_knn_fit_ot,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_tr_knn_ot<-mean((Train_D_Dataset$Otro-y_tr_pred_knn_ot)^2) # calcula el mse de entrenamiento
RMSE_tr_knn_ot = sqrt(mse_tr_knn_ot)
mse_tr_knn_ot## [1] 22.38182RMSE_tr_knn_ot## [1] 4.730943y_test_pred_knn_ot<-predict(caret_knn_fit_ot,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_test_knn_ot<-mean((Test_D_Dataset$Otro-y_test_pred_knn_ot)^2) # calcula el mse de entrenamiento
RMSE_test_knn_ot = sqrt(mse_test_knn_ot)
mse_test_knn_ot## [1] 24.70889RMSE_test_knn_ot## [1] 4.970804plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Otro,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_knn_ot,
name='Modelo knn',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Otros Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))Resumen Modelos KNN para los diferentes tipos de accidente
- TOTAL ACCIDENTES: MSE = 629.9185512 - RMSE = 629.9185512
2. REGRESION LINEAL
Total Accidentes
head(Train_D_Dataset)## FECHA Atropello Caida_Ocupante Choque Otro Volcamiento Lunes martes
## 1 2014-01-01 13 7 35 18 1 0 0
## 2 2014-01-02 12 7 43 9 1 0 0
## 3 2014-01-03 7 5 67 13 1 0 0
## 4 2014-01-04 11 7 40 9 1 0 0
## 5 2014-01-05 6 5 43 10 3 0 0
## 6 2014-01-06 4 5 23 9 2 1 0
## miercoles jueves viernes sabado domingo Enero Febrero Marzo Abril Mayo
## 1 1 0 0 0 0 1 0 0 0 0
## 2 0 1 0 0 0 1 0 0 0 0
## 3 0 0 1 0 0 1 0 0 0 0
## 4 0 0 0 1 0 1 0 0 0 0
## 5 0 0 0 0 1 1 0 0 0 0
## 6 0 0 0 0 0 1 0 0 0 0
## Junio Julio Agosto Septiembre Octubre Noviembre Diciembre Feriado
## 1 0 0 0 0 0 0 0 1
## 2 0 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 0 0
## 4 0 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0 1
## Semana.Santa Prima Mujer Padre Madre AmoryAmistad Viernes_Antes_Puente
## 1 0 0 0 0 0 0 0
## 2 0 0 0 0 0 0 0
## 3 0 0 0 0 0 0 1
## 4 0 0 0 0 0 0 0
## 5 0 0 0 0 0 0 0
## 6 0 0 0 0 0 0 0
## Quincena Viernes_Desp_Qincena ANO SEMANA DIA DIA_SEMANA MES
## 1 0 0 2014 01 01 miércoles 01
## 2 0 0 2014 01 02 jueves 01
## 3 0 0 2014 01 03 viernes 01
## 4 0 0 2014 01 04 sábado 01
## 5 0 0 2014 01 05 domingo 01
## 6 0 0 2014 02 06 lunes 01
## Total_Accidentes Accidentes_Graves
## 1 74 21
## 2 72 20
## 3 93 13
## 4 68 19
## 5 67 14
## 6 43 11trcntrl = trainControl(method="cv", number=10)
caret_lm_fit = caret::train(Total_Accidentes~SEMANA+DIA_SEMANA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente, data=Train_D_Dataset,
method = "lm", trControl = trcntrl,
preProcess=c("center", "scale"),
tuneLength = 10)summary(caret_lm_fit)##
## Call:
## lm(formula = .outcome ~ ., data = dat)
##
## Residuals:
## Min 1Q Median 3Q Max
## -56.162 -10.047 -0.588 8.959 69.282
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 115.7276 0.4078 283.784 < 2e-16 ***
## SEMANA02 1.2376 0.5835 2.121 0.03410 *
## SEMANA03 3.4399 0.5844 5.886 4.94e-09 ***
## SEMANA04 3.5403 0.5849 6.053 1.82e-09 ***
## SEMANA05 4.1744 0.5846 7.140 1.49e-12 ***
## SEMANA06 4.8285 0.5846 8.260 3.35e-16 ***
## SEMANA07 5.1933 0.5848 8.880 < 2e-16 ***
## SEMANA08 4.7466 0.5844 8.122 9.99e-16 ***
## SEMANA09 4.6656 0.5849 7.977 3.11e-15 ***
## SEMANA10 5.6857 0.5846 9.726 < 2e-16 ***
## SEMANA11 5.5909 0.5842 9.571 < 2e-16 ***
## SEMANA12 4.4708 0.5919 7.554 7.60e-14 ***
## SEMANA13 5.0444 0.5848 8.627 < 2e-16 ***
## SEMANA14 5.5155 0.5996 9.199 < 2e-16 ***
## SEMANA15 5.3791 0.6000 8.966 < 2e-16 ***
## SEMANA16 4.0457 0.5940 6.811 1.44e-11 ***
## SEMANA17 5.6406 0.5837 9.664 < 2e-16 ***
## SEMANA18 5.4263 0.5835 9.300 < 2e-16 ***
## SEMANA19 5.1669 0.5845 8.840 < 2e-16 ***
## SEMANA20 5.2001 0.5838 8.907 < 2e-16 ***
## SEMANA21 4.9700 0.5829 8.527 < 2e-16 ***
## SEMANA22 4.3113 0.5830 7.395 2.43e-13 ***
## SEMANA23 5.2293 0.5831 8.968 < 2e-16 ***
## SEMANA24 4.7994 0.5829 8.234 4.11e-16 ***
## SEMANA25 3.9004 0.5825 6.696 3.10e-11 ***
## SEMANA26 3.0130 0.5825 5.172 2.65e-07 ***
## SEMANA27 4.2672 0.5841 7.305 4.64e-13 ***
## SEMANA28 4.6929 0.5849 8.024 2.15e-15 ***
## SEMANA29 5.2255 0.5835 8.956 < 2e-16 ***
## SEMANA30 5.2697 0.5845 9.016 < 2e-16 ***
## SEMANA31 6.6748 0.5835 11.440 < 2e-16 ***
## SEMANA32 6.0952 0.5832 10.450 < 2e-16 ***
## SEMANA33 5.3466 0.5829 9.172 < 2e-16 ***
## SEMANA34 4.5076 0.5839 7.720 2.21e-14 ***
## SEMANA35 5.2800 0.5845 9.033 < 2e-16 ***
## SEMANA36 4.8264 0.5843 8.260 3.35e-16 ***
## SEMANA37 5.8110 0.5845 9.941 < 2e-16 ***
## SEMANA38 5.8040 0.5845 9.931 < 2e-16 ***
## SEMANA39 4.5949 0.5849 7.856 7.84e-15 ***
## SEMANA40 5.7074 0.5847 9.761 < 2e-16 ***
## SEMANA41 3.7066 0.5832 6.355 2.81e-10 ***
## SEMANA42 4.7340 0.5844 8.100 1.18e-15 ***
## SEMANA43 4.8565 0.5849 8.302 2.38e-16 ***
## SEMANA44 4.6772 0.5833 8.019 2.24e-15 ***
## SEMANA45 3.9045 0.5828 6.700 3.02e-11 ***
## SEMANA46 4.8501 0.5830 8.319 < 2e-16 ***
## SEMANA47 4.6346 0.5839 7.937 4.21e-15 ***
## SEMANA48 4.6285 0.5845 7.918 4.88e-15 ***
## SEMANA49 4.9809 0.5830 8.543 < 2e-16 ***
## SEMANA50 5.1014 0.5840 8.735 < 2e-16 ***
## SEMANA51 5.8308 0.5833 9.996 < 2e-16 ***
## SEMANA52 2.8152 0.5837 4.823 1.57e-06 ***
## SEMANA53 0.1876 0.4593 0.408 0.68301
## DIA_SEMANAjueves 15.3319 0.5345 28.686 < 2e-16 ***
## DIA_SEMANAlunes 15.6617 0.5459 28.687 < 2e-16 ***
## DIA_SEMANAmartes 16.2709 0.5352 30.404 < 2e-16 ***
## DIA_SEMANAmiércoles 15.2980 0.5354 28.575 < 2e-16 ***
## DIA_SEMANAsábado 12.6912 0.5359 23.681 < 2e-16 ***
## DIA_SEMANAviernes 17.2568 0.6806 25.356 < 2e-16 ***
## Semana.Santa -1.3413 0.4942 -2.714 0.00672 **
## Feriado -10.1955 0.4632 -22.010 < 2e-16 ***
## Quincena -0.3212 0.4165 -0.771 0.44070
## Viernes_Desp_Qincena 0.1768 0.5527 0.320 0.74911
## Viernes_Antes_Puente 0.4475 0.4718 0.949 0.34300
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 15.59 on 1397 degrees of freedom
## Multiple R-squared: 0.6585, Adjusted R-squared: 0.6431
## F-statistic: 42.76 on 63 and 1397 DF, p-value: < 2.2e-16caret_lm_fit## Linear Regression
##
## 1461 samples
## 7 predictor
##
## Pre-processing: centered (63), scaled (63)
## Resampling: Cross-Validated (10 fold)
## Summary of sample sizes: 1316, 1314, 1314, 1316, 1315, 1314, ...
## Resampling results:
##
## RMSE Rsquared MAE
## 15.92204 0.6314618 12.31408
##
## Tuning parameter 'intercept' was held constant at a value of TRUECalculo MSE y RMSE para los datos de entrenamiento
y_tr_pred_lm<-predict(caret_lm_fit,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])mse_tr_lm<-mean((Train_D_Dataset$Total_Accidentes-y_tr_pred_lm)^2) # calcula el mse de entrenamiento
RMSE_tr_lm = sqrt(mse_tr_lm)
mse_tr_lm## [1] 232.3244RMSE_tr_lm## [1] 15.24219y_test_pred_lm<-predict(caret_lm_fit,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_test_lm<-mean((Test_D_Dataset$Total_Accidentes-y_test_pred_lm)^2) # calcula el mse de entrenamiento
RMSE_test_lm = sqrt(mse_test_lm)
mse_test_lm## [1] 254.0824RMSE_test_lm## [1] 15.93996PRedicion en la muestra
plot_ly (data=Train_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_tr_pred_lm,
name='Modelo lm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))Grafica serie 2018
plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_lm,
name='Modelo lm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))Choques
trcntrl = trainControl(method="cv", number=10)
caret_lm_fit_ch = caret::train(Choque~SEMANA+DIA_SEMANA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente, data=Train_D_Dataset,
method = "lm", trControl = trcntrl,
preProcess=c("center", "scale"),
tuneLength = 10)summary(caret_lm_fit_ch)##
## Call:
## lm(formula = .outcome ~ ., data = dat)
##
## Residuals:
## Min 1Q Median 3Q Max
## -45.118 -7.832 -0.475 7.333 44.379
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 77.50376 0.31378 247.000 < 2e-16 ***
## SEMANA02 1.26133 0.44900 2.809 0.00504 **
## SEMANA03 2.64864 0.44966 5.890 4.82e-09 ***
## SEMANA04 3.00263 0.45001 6.672 3.62e-11 ***
## SEMANA05 3.26434 0.44984 7.257 6.56e-13 ***
## SEMANA06 3.35852 0.44980 7.467 1.44e-13 ***
## SEMANA07 4.31915 0.44999 9.598 < 2e-16 ***
## SEMANA08 3.89218 0.44968 8.656 < 2e-16 ***
## SEMANA09 3.72621 0.45006 8.279 2.86e-16 ***
## SEMANA10 4.15694 0.44980 9.242 < 2e-16 ***
## SEMANA11 4.07270 0.44948 9.061 < 2e-16 ***
## SEMANA12 3.37787 0.45540 7.417 2.07e-13 ***
## SEMANA13 3.68188 0.44993 8.183 6.16e-16 ***
## SEMANA14 4.27341 0.46136 9.263 < 2e-16 ***
## SEMANA15 4.38193 0.46164 9.492 < 2e-16 ***
## SEMANA16 3.19976 0.45704 7.001 3.93e-12 ***
## SEMANA17 4.54173 0.44911 10.113 < 2e-16 ***
## SEMANA18 4.21887 0.44893 9.398 < 2e-16 ***
## SEMANA19 4.10791 0.44975 9.134 < 2e-16 ***
## SEMANA20 4.06900 0.44920 9.058 < 2e-16 ***
## SEMANA21 3.51132 0.44847 7.829 9.62e-15 ***
## SEMANA22 3.24565 0.44858 7.235 7.63e-13 ***
## SEMANA23 3.76270 0.44865 8.387 < 2e-16 ***
## SEMANA24 3.64271 0.44848 8.122 9.95e-16 ***
## SEMANA25 2.97196 0.44820 6.631 4.76e-11 ***
## SEMANA26 2.58447 0.44822 5.766 9.98e-09 ***
## SEMANA27 3.37844 0.44946 7.517 1.00e-13 ***
## SEMANA28 3.46450 0.45002 7.698 2.59e-14 ***
## SEMANA29 3.92987 0.44893 8.754 < 2e-16 ***
## SEMANA30 4.07362 0.44975 9.058 < 2e-16 ***
## SEMANA31 4.94896 0.44895 11.023 < 2e-16 ***
## SEMANA32 4.66251 0.44878 10.389 < 2e-16 ***
## SEMANA33 3.96387 0.44852 8.838 < 2e-16 ***
## SEMANA34 3.39716 0.44928 7.561 7.19e-14 ***
## SEMANA35 3.39685 0.44978 7.552 7.69e-14 ***
## SEMANA36 3.79075 0.44962 8.431 < 2e-16 ***
## SEMANA37 4.23817 0.44976 9.423 < 2e-16 ***
## SEMANA38 4.36502 0.44971 9.706 < 2e-16 ***
## SEMANA39 3.41552 0.45002 7.590 5.83e-14 ***
## SEMANA40 4.51661 0.44992 10.039 < 2e-16 ***
## SEMANA41 3.17107 0.44876 7.066 2.50e-12 ***
## SEMANA42 3.49085 0.44967 7.763 1.59e-14 ***
## SEMANA43 3.71145 0.45008 8.246 3.73e-16 ***
## SEMANA44 3.91307 0.44881 8.719 < 2e-16 ***
## SEMANA45 2.95727 0.44842 6.595 6.02e-11 ***
## SEMANA46 3.90202 0.44858 8.699 < 2e-16 ***
## SEMANA47 3.49686 0.44928 7.783 1.37e-14 ***
## SEMANA48 3.88178 0.44978 8.630 < 2e-16 ***
## SEMANA49 4.22457 0.44860 9.417 < 2e-16 ***
## SEMANA50 4.45513 0.44937 9.914 < 2e-16 ***
## SEMANA51 4.60465 0.44882 10.259 < 2e-16 ***
## SEMANA52 1.99434 0.44915 4.440 9.69e-06 ***
## SEMANA53 0.41755 0.35344 1.181 0.23765
## DIA_SEMANAjueves 12.29178 0.41124 29.889 < 2e-16 ***
## DIA_SEMANAlunes 12.86427 0.42008 30.624 < 2e-16 ***
## DIA_SEMANAmartes 13.56922 0.41178 32.953 < 2e-16 ***
## DIA_SEMANAmiércoles 12.56514 0.41193 30.503 < 2e-16 ***
## DIA_SEMANAsábado 10.60223 0.41236 25.711 < 2e-16 ***
## DIA_SEMANAviernes 14.50901 0.52367 27.706 < 2e-16 ***
## Semana.Santa -0.89104 0.38024 -2.343 0.01925 *
## Feriado -8.51712 0.35643 -23.896 < 2e-16 ***
## Quincena -0.56810 0.32046 -1.773 0.07648 .
## Viernes_Desp_Qincena 0.01274 0.42529 0.030 0.97611
## Viernes_Antes_Puente 0.47114 0.36301 1.298 0.19455
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 11.99 on 1397 degrees of freedom
## Multiple R-squared: 0.6841, Adjusted R-squared: 0.6698
## F-statistic: 48.01 on 63 and 1397 DF, p-value: < 2.2e-16caret_lm_fit_ch## Linear Regression
##
## 1461 samples
## 7 predictor
##
## Pre-processing: centered (63), scaled (63)
## Resampling: Cross-Validated (10 fold)
## Summary of sample sizes: 1317, 1314, 1314, 1315, 1316, 1314, ...
## Resampling results:
##
## RMSE Rsquared MAE
## 12.33502 0.6547476 9.712046
##
## Tuning parameter 'intercept' was held constant at a value of TRUEy_tr_pred_lm_ch<-predict(caret_lm_fit_ch,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])mse_tr_lm_ch<-mean((Train_D_Dataset$Choque-y_tr_pred_lm_ch)^2) # calcula el mse de entrenamiento
RMSE_tr_lm_ch = sqrt(mse_tr_lm_ch)
mse_tr_lm_ch## [1] 137.5459RMSE_tr_lm_ch## [1] 11.728y_test_pred_lm_ch<-predict(caret_lm_fit_ch,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_test_lm_ch<-mean((Test_D_Dataset$Choque-y_test_pred_lm_ch)^2) # calcula el mse de entrenamiento
RMSE_test_lm_ch = sqrt(mse_test_lm_ch)
mse_test_lm_ch## [1] 149.9768RMSE_test_lm_ch## [1] 12.2465PRedicion en la muestra
plot_ly (data=Train_D_Dataset,
x = ~FECHA,
y = ~Choque,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_tr_pred_lm_ch,
name='Modelo lm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))Grafica serie 2018
plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Choque,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_lm_ch,
name='Modelo lm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))3. MODELO LINEAL GENERALIZADO
glm_fit<-glm(Total_Accidentes~SEMANA+DIA_SEMANA+DIA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente, data=Train_D_Dataset, family = "poisson")
summary(glm_fit)##
## Call:
## glm(formula = Total_Accidentes ~ SEMANA + DIA_SEMANA + DIA +
## Semana.Santa + Feriado + Quincena + Viernes_Desp_Qincena +
## Viernes_Antes_Puente, family = "poisson", data = Train_D_Dataset)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -5.6815 -0.9401 -0.0613 0.8115 5.9142
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 4.052e+00 2.712e-02 149.378 < 2e-16 ***
## SEMANA02 8.972e-02 3.065e-02 2.927 0.00342 **
## SEMANA03 2.439e-01 3.041e-02 8.019 1.07e-15 ***
## SEMANA04 2.780e-01 2.995e-02 9.282 < 2e-16 ***
## SEMANA05 3.272e-01 2.860e-02 11.438 < 2e-16 ***
## SEMANA06 3.515e-01 2.832e-02 12.411 < 2e-16 ***
## SEMANA07 3.583e-01 2.951e-02 12.140 < 2e-16 ***
## SEMANA08 3.392e-01 2.981e-02 11.378 < 2e-16 ***
## SEMANA09 3.517e-01 2.891e-02 12.166 < 2e-16 ***
## SEMANA10 4.004e-01 2.805e-02 14.275 < 2e-16 ***
## SEMANA11 3.785e-01 2.947e-02 12.842 < 2e-16 ***
## SEMANA12 3.211e-01 3.063e-02 10.486 < 2e-16 ***
## SEMANA13 3.820e-01 2.918e-02 13.093 < 2e-16 ***
## SEMANA14 3.992e-01 2.867e-02 13.927 < 2e-16 ***
## SEMANA15 3.768e-01 2.979e-02 12.647 < 2e-16 ***
## SEMANA16 2.818e-01 3.076e-02 9.160 < 2e-16 ***
## SEMANA17 4.068e-01 2.930e-02 13.883 < 2e-16 ***
## SEMANA18 4.040e-01 2.811e-02 14.374 < 2e-16 ***
## SEMANA19 3.713e-01 2.867e-02 12.952 < 2e-16 ***
## SEMANA20 3.524e-01 2.970e-02 11.867 < 2e-16 ***
## SEMANA21 3.661e-01 2.968e-02 12.336 < 2e-16 ***
## SEMANA22 3.429e-01 2.890e-02 11.864 < 2e-16 ***
## SEMANA23 3.764e-01 2.813e-02 13.383 < 2e-16 ***
## SEMANA24 3.425e-01 2.956e-02 11.588 < 2e-16 ***
## SEMANA25 2.804e-01 3.030e-02 9.256 < 2e-16 ***
## SEMANA26 2.480e-01 2.986e-02 8.307 < 2e-16 ***
## SEMANA27 3.247e-01 2.855e-02 11.373 < 2e-16 ***
## SEMANA28 3.362e-01 2.919e-02 11.515 < 2e-16 ***
## SEMANA29 3.545e-01 2.998e-02 11.825 < 2e-16 ***
## SEMANA30 3.929e-01 2.942e-02 13.356 < 2e-16 ***
## SEMANA31 4.705e-01 2.775e-02 16.956 < 2e-16 ***
## SEMANA32 4.352e-01 2.827e-02 15.394 < 2e-16 ***
## SEMANA33 3.644e-01 2.958e-02 12.320 < 2e-16 ***
## SEMANA34 3.359e-01 3.013e-02 11.148 < 2e-16 ***
## SEMANA35 3.997e-01 2.872e-02 13.918 < 2e-16 ***
## SEMANA36 3.519e-01 2.806e-02 12.543 < 2e-16 ***
## SEMANA37 4.033e-01 2.905e-02 13.884 < 2e-16 ***
## SEMANA38 3.913e-01 2.965e-02 13.200 < 2e-16 ***
## SEMANA39 3.497e-01 2.941e-02 11.893 < 2e-16 ***
## SEMANA40 4.119e-01 2.779e-02 14.819 < 2e-16 ***
## SEMANA41 2.748e-01 2.923e-02 9.403 < 2e-16 ***
## SEMANA42 3.301e-01 3.025e-02 10.911 < 2e-16 ***
## SEMANA43 3.611e-01 2.960e-02 12.201 < 2e-16 ***
## SEMANA44 3.608e-01 2.844e-02 12.686 < 2e-16 ***
## SEMANA45 2.949e-01 2.892e-02 10.198 < 2e-16 ***
## SEMANA46 3.405e-01 2.971e-02 11.460 < 2e-16 ***
## SEMANA47 3.351e-01 3.008e-02 11.142 < 2e-16 ***
## SEMANA48 3.542e-01 2.901e-02 12.211 < 2e-16 ***
## SEMANA49 3.637e-01 2.812e-02 12.936 < 2e-16 ***
## SEMANA50 3.649e-01 2.943e-02 12.401 < 2e-16 ***
## SEMANA51 3.893e-01 2.967e-02 13.119 < 2e-16 ***
## SEMANA52 2.298e-01 3.041e-02 7.559 4.08e-14 ***
## SEMANA53 4.368e-02 4.806e-02 0.909 0.36343
## DIA_SEMANAjueves 4.468e-01 1.007e-02 44.387 < 2e-16 ***
## DIA_SEMANAlunes 4.601e-01 1.030e-02 44.647 < 2e-16 ***
## DIA_SEMANAmartes 4.675e-01 1.002e-02 46.674 < 2e-16 ***
## DIA_SEMANAmiércoles 4.453e-01 1.005e-02 44.304 < 2e-16 ***
## DIA_SEMANAsábado 3.836e-01 1.018e-02 37.682 < 2e-16 ***
## DIA_SEMANAviernes 4.978e-01 1.228e-02 40.533 < 2e-16 ***
## DIA02 -6.085e-02 1.942e-02 -3.134 0.00172 **
## DIA03 -2.768e-02 1.942e-02 -1.425 0.15407
## DIA04 -3.967e-02 1.962e-02 -2.022 0.04317 *
## DIA05 3.952e-03 1.970e-02 0.201 0.84104
## DIA06 -1.417e-02 2.015e-02 -0.703 0.48188
## DIA07 2.947e-02 2.034e-02 1.449 0.14739
## DIA08 -2.281e-02 2.095e-02 -1.089 0.27608
## DIA09 -4.209e-02 2.113e-02 -1.992 0.04634 *
## DIA10 7.543e-04 2.114e-02 0.036 0.97154
## DIA11 -1.591e-02 2.149e-02 -0.740 0.45924
## DIA12 -3.117e-02 2.182e-02 -1.428 0.15315
## DIA13 -3.675e-02 2.197e-02 -1.673 0.09438 .
## DIA14 -1.467e-02 2.190e-02 -0.670 0.50302
## DIA15 3.408e-03 3.421e-02 0.100 0.92064
## DIA16 2.433e-02 2.173e-02 1.120 0.26276
## DIA17 2.914e-02 2.172e-02 1.342 0.17975
## DIA18 3.754e-02 2.171e-02 1.729 0.08383 .
## DIA19 -6.384e-03 2.182e-02 -0.293 0.76986
## DIA20 -6.073e-05 2.174e-02 -0.003 0.99777
## DIA21 5.211e-03 2.139e-02 0.244 0.80752
## DIA22 -3.261e-02 2.142e-02 -1.522 0.12792
## DIA23 -1.240e-02 2.113e-02 -0.587 0.55721
## DIA24 -4.334e-02 2.091e-02 -2.072 0.03823 *
## DIA25 -2.404e-02 2.068e-02 -1.162 0.24509
## DIA26 -4.959e-02 2.052e-02 -2.417 0.01564 *
## DIA27 -4.147e-02 2.000e-02 -2.073 0.03818 *
## DIA28 -1.227e-02 1.965e-02 -0.624 0.53231
## DIA29 -3.712e-02 2.001e-02 -1.856 0.06352 .
## DIA30 -4.994e-02 2.248e-02 -2.221 0.02632 *
## DIA31 -7.889e-02 3.597e-02 -2.193 0.02832 *
## Semana.Santa -1.017e-01 2.094e-02 -4.857 1.19e-06 ***
## Feriado -5.012e-01 1.504e-02 -33.315 < 2e-16 ***
## Quincena 2.597e-03 2.629e-02 0.099 0.92131
## Viernes_Desp_Qincena -1.226e-02 1.351e-02 -0.907 0.36420
## Viernes_Antes_Puente 2.202e-02 1.559e-02 1.412 0.15788
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for poisson family taken to be 1)
##
## Null deviance: 9170.5 on 1460 degrees of freedom
## Residual deviance: 2791.2 on 1367 degrees of freedom
## AIC: 12564
##
## Number of Fisher Scoring iterations: 4glm_fit##
## Call: glm(formula = Total_Accidentes ~ SEMANA + DIA_SEMANA + DIA +
## Semana.Santa + Feriado + Quincena + Viernes_Desp_Qincena +
## Viernes_Antes_Puente, family = "poisson", data = Train_D_Dataset)
##
## Coefficients:
## (Intercept) SEMANA02 SEMANA03
## 4.052e+00 8.972e-02 2.439e-01
## SEMANA04 SEMANA05 SEMANA06
## 2.780e-01 3.272e-01 3.515e-01
## SEMANA07 SEMANA08 SEMANA09
## 3.583e-01 3.392e-01 3.517e-01
## SEMANA10 SEMANA11 SEMANA12
## 4.004e-01 3.785e-01 3.211e-01
## SEMANA13 SEMANA14 SEMANA15
## 3.820e-01 3.992e-01 3.768e-01
## SEMANA16 SEMANA17 SEMANA18
## 2.818e-01 4.068e-01 4.040e-01
## SEMANA19 SEMANA20 SEMANA21
## 3.713e-01 3.524e-01 3.661e-01
## SEMANA22 SEMANA23 SEMANA24
## 3.429e-01 3.764e-01 3.425e-01
## SEMANA25 SEMANA26 SEMANA27
## 2.804e-01 2.480e-01 3.247e-01
## SEMANA28 SEMANA29 SEMANA30
## 3.362e-01 3.545e-01 3.929e-01
## SEMANA31 SEMANA32 SEMANA33
## 4.705e-01 4.352e-01 3.644e-01
## SEMANA34 SEMANA35 SEMANA36
## 3.359e-01 3.997e-01 3.519e-01
## SEMANA37 SEMANA38 SEMANA39
## 4.033e-01 3.913e-01 3.497e-01
## SEMANA40 SEMANA41 SEMANA42
## 4.119e-01 2.748e-01 3.301e-01
## SEMANA43 SEMANA44 SEMANA45
## 3.611e-01 3.608e-01 2.949e-01
## SEMANA46 SEMANA47 SEMANA48
## 3.405e-01 3.351e-01 3.542e-01
## SEMANA49 SEMANA50 SEMANA51
## 3.637e-01 3.649e-01 3.893e-01
## SEMANA52 SEMANA53 DIA_SEMANAjueves
## 2.298e-01 4.368e-02 4.468e-01
## DIA_SEMANAlunes DIA_SEMANAmartes DIA_SEMANAmiércoles
## 4.601e-01 4.675e-01 4.453e-01
## DIA_SEMANAsábado DIA_SEMANAviernes DIA02
## 3.836e-01 4.978e-01 -6.085e-02
## DIA03 DIA04 DIA05
## -2.768e-02 -3.967e-02 3.952e-03
## DIA06 DIA07 DIA08
## -1.417e-02 2.947e-02 -2.281e-02
## DIA09 DIA10 DIA11
## -4.209e-02 7.542e-04 -1.591e-02
## DIA12 DIA13 DIA14
## -3.117e-02 -3.675e-02 -1.467e-02
## DIA15 DIA16 DIA17
## 3.408e-03 2.433e-02 2.914e-02
## DIA18 DIA19 DIA20
## 3.754e-02 -6.384e-03 -6.073e-05
## DIA21 DIA22 DIA23
## 5.211e-03 -3.261e-02 -1.240e-02
## DIA24 DIA25 DIA26
## -4.334e-02 -2.404e-02 -4.959e-02
## DIA27 DIA28 DIA29
## -4.147e-02 -1.227e-02 -3.712e-02
## DIA30 DIA31 Semana.Santa
## -4.994e-02 -7.889e-02 -1.017e-01
## Feriado Quincena Viernes_Desp_Qincena
## -5.012e-01 2.597e-03 -1.226e-02
## Viernes_Antes_Puente
## 2.202e-02
##
## Degrees of Freedom: 1460 Total (i.e. Null); 1367 Residual
## Null Deviance: 9171
## Residual Deviance: 2791 AIC: 12560y_tr_pred_glm<-predict(glm_fit,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","DIA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")],type="response")
mse_tr_glm<-mean((Train_D_Dataset$Total_Accidentes-y_tr_pred_glm)^2) # calcula el mse de entrenamiento
RMSE_tr_glm = sqrt(mse_tr_glm)
mse_tr_glm## [1] 211.596RMSE_tr_glm## [1] 14.54634y_test_pred_glm<-predict(glm_fit,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","DIA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")],type="response")
mse_test_glm<-mean((Train_D_Dataset$Total_Accidentes-y_test_pred_glm)^2) # calcula el mse de entrenamiento## Warning in Train_D_Dataset$Total_Accidentes - y_test_pred_glm: longitud de
## objeto mayor no es múltiplo de la longitud de uno menorRMSE_test_glm = sqrt(mse_test_glm)
mse_test_glm## [1] 1101.145RMSE_test_glm## [1] 33.18351plot_ly (data=Train_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_tr_pred_glm,
name='Modelo glm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_glm,
name='Modelo glm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))4. ARBOLES DE REGRESION
trcntrl = trainControl(method="cv", number=10)
caret_tree_fit = caret::train(Total_Accidentes~SEMANA+DIA_SEMANA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente,data=Train_D_Dataset,
method = "rpart", trControl = trcntrl,
parms = list(split = "gini"),
preProcess=c("center", "scale"),
tuneLength = 10)## Warning in nominalTrainWorkflow(x = x, y = y, wts = weights, info =
## trainInfo, : There were missing values in resampled performance measures.caret_tree_fit## CART
##
## 1461 samples
## 7 predictor
##
## Pre-processing: centered (63), scaled (63)
## Resampling: Cross-Validated (10 fold)
## Summary of sample sizes: 1316, 1316, 1315, 1315, 1316, 1314, ...
## Resampling results across tuning parameters:
##
## cp RMSE Rsquared MAE
## 6.424688e-05 16.80676 0.5855434 12.84698
## 1.890736e-04 16.80781 0.5854719 12.84759
## 2.089483e-04 16.80854 0.5854612 12.84841
## 6.408283e-04 16.80061 0.5859644 12.85642
## 8.458601e-04 16.76004 0.5879283 12.85201
## 1.709459e-03 16.79462 0.5866256 12.85094
## 2.310304e-03 16.81679 0.5856843 12.86666
## 3.426222e-03 16.82144 0.5855254 12.86090
## 6.807466e-02 19.76936 0.4057811 15.24499
## 1.793061e-01 25.37186 0.1288797 19.91892
##
## RMSE was used to select the optimal model using the smallest value.
## The final value used for the model was cp = 0.0008458601.y_tr_pred_tree<-predict(caret_tree_fit,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_tr_tree<-mean((Train_D_Dataset$Total_Accidentes-y_tr_pred_tree)^2) # calcula el mse de entrenamiento
RMSE_tr_tree = sqrt(mse_tr_tree)
mse_tr_tree## [1] 275.4052RMSE_tr_tree## [1] 16.59534y_test_pred_tree<-predict(caret_tree_fit,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_test_tree<-mean((Test_D_Dataset$Total_Accidentes-y_test_pred_tree)^2) # calcula el mse de entrenamiento
RMSE_test_tree = sqrt(mse_test_tree)
mse_test_tree## [1] 269.4619RMSE_test_tree## [1] 16.41529plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_tree,
name='Modelo lm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))plot_ly (data=Train_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_tr_pred_tree,
name='Modelo lm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))5. BOSQUE ALEATORIO
trcntrl = trainControl(method="cv", number=10)
caret_rf_fit = caret::train(Total_Accidentes~SEMANA+DIA_SEMANA+Semana.Santa+Feriado+Quincena+Viernes_Desp_Qincena+Viernes_Antes_Puente, data=Train_D_Dataset,
method = "rf", trControl = trcntrl,
prox=TRUE,allowParallel=TRUE)summary(caret_rf_fit)## Length Class Mode
## call 6 -none- call
## type 1 -none- character
## predicted 1461 -none- numeric
## mse 500 -none- numeric
## rsq 500 -none- numeric
## oob.times 1461 -none- numeric
## importance 63 -none- numeric
## importanceSD 0 -none- NULL
## localImportance 0 -none- NULL
## proximity 2134521 -none- numeric
## ntree 1 -none- numeric
## mtry 1 -none- numeric
## forest 11 -none- list
## coefs 0 -none- NULL
## y 1461 -none- numeric
## test 0 -none- NULL
## inbag 0 -none- NULL
## xNames 63 -none- character
## problemType 1 -none- character
## tuneValue 1 data.frame list
## obsLevels 1 -none- logical
## param 2 -none- listcaret_rf_fit## Random Forest
##
## 1461 samples
## 7 predictor
##
## No pre-processing
## Resampling: Cross-Validated (10 fold)
## Summary of sample sizes: 1315, 1314, 1313, 1316, 1315, 1315, ...
## Resampling results across tuning parameters:
##
## mtry RMSE Rsquared MAE
## 2 22.59901 0.3847643 17.97227
## 32 17.81764 0.5517189 13.69226
## 63 17.95390 0.5488941 13.79388
##
## RMSE was used to select the optimal model using the smallest value.
## The final value used for the model was mtry = 32.plot(caret_rf_fit)
y_tr_pred_rf<-predict(caret_rf_fit,Train_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_tr_rf<-mean((Train_D_Dataset$Total_Accidentes-y_tr_pred_rf)^2) # calcula el mse de entrenamiento
RMSE_tr_rf = sqrt(mse_tr_rf)
mse_tr_rf## [1] 152.6926RMSE_tr_rf## [1] 12.35689y_test_pred_rf<-predict(caret_rf_fit,Test_D_Dataset[,c("SEMANA","DIA_SEMANA","Semana.Santa","Feriado","Quincena","Viernes_Desp_Qincena","Viernes_Antes_Puente")])
mse_test_rf<-mean((Test_D_Dataset$Total_Accidentes-y_test_pred_rf)^2) # calcula el mse de entrenamiento
RMSE_test_rf = sqrt(mse_test_rf)
mse_test_rf## [1] 284.6565RMSE_test_rf## [1] 16.87177plot_ly (data=Test_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_test_pred_rf,
name='Modelo lm',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))plot_ly (data=Train_D_Dataset,
x = ~FECHA,
y = ~Total_Accidentes,
type = "scatter" ,mode = "lines",
name='Real',
line=list(width=1,color='rgb(205, 12, 24)'))%>%
add_trace(y= ~y_tr_pred_rf,
name='Modelo rf',
line=list(width=1,color='rgb(22, 96, 167)'))%>%
layout(title='Total Accidentes',
xaxis=list(title="Fecha"),
yaxis=list(title="Accidentes"),
legend = list(x = 0.75, y = 0.9))